Drones have become increasingly popular in precision agriculture due to their ability to collect valuable data quickly and efficiently. One of the major aspects of precision agriculture is to estimate fraction crop cover at an early stage. This paper develops an approach using fine-tuned Machine Learning (ML) YOLO models to extract crop fields only from drone imagery and mask all other objects. A combination of Color Space Models (CSM) is used to extract fraction crop cover at an early stage. The approach was developed for extracting information with less processing complexity as drones/UAVs have limited processing and power capabilities. The primary objective of this study is to identify low-density crop areas and barren land within crop fields during the early stages of crop growth using CSM. Otsu and Max Entropy thresholding techniques are analysed to obtain mask information of targeted area. Morphological open and close operations are used to get desired size patches of sparse or no crop location. The study suggests Otsu thresholding as an adaptive thresholding method as its results are adequate compared to ground truth. Different filter size results are also compared as filter size determines the minimum patch size identified on the ground. The finetuned ML model extracts the object of interest. The color space model works well when applied to that single object.

It is still a challenge for existing DNN based models to synergistically exploit the spatial, temporal, and especially spectral information of a crop present in multi-sensor time series (MSTS) remote sensing (RS) images and provide accurate crop classification while keeping the generalization ability of DNN models high. This imbalance requires investigation and demands novel CNN and RNN model-based approaches that can address the issue. The novel models proposed in this study involve the concepts of permuted localized spectral convolutions, localized spatial convolutions, and bi-directional recurrent units. The permuted spectral band stacking strategy is explored in this study to strengthen the influence of the spectral information. Overall, 6 models are proposed namely; Perm-1D-CNN, Perm-3D-CNN, Perm-RNN, Perm-1D-CRNN, Perm-2D-CRNN, and Perm-3D-CRNN. The qualitative and quantitative assessments reflect the higher generalization ability of the Perm-3D-CRNN along with its high classification accuracy. Also, the impact of spectral band permutations and localized spectral convolutions on the performance of DNN models is significant toward improved generalization.

Nowadays, in through-wall imaging, the simultaneous detection of the dielectric contrast materials like wood and metal placed together behind the wall is quite a challenging task due to the strong reflections. This adds on to the limitation when targets are placed in complex environment like window that also gives the false target information. In this paper, a methodology for the detection of low and high dielectric target is developed and explored. The experimental data is collected in the through-wall imaging scenario for the contrast dielectric targets detection with window effect on the other side of the wall. Further, optimum rank thresholding is applied using the opt-shrink algorithm with the exploitation of noise subspace in singular value decomposition (SVD) for the detection of low dielectric material. A suitable agreement with the results of targets detection is obtained as per the scenario in complex environment with the proposed methodology.

Accurate information on spatial distribution of crop and vegetation indices for crop health monitoring is important for precision agriculture monitoring. However, freely available multispectral satellite images and unmanned aerial vehicle based multispectral images provides great opportunities for crop area estimation and extraction of vegetation indices. An object-based image analysis is better than pixel-based analysis because it is used statistical, geometrical, and topographic feature of the objects. Therefore, this paper presents an object-based image analysis of multispectral satellite and drone images for crop area estimation and extraction of vegetation indices for precision agriculture monitoring. Object segmentation, feature extraction, and classification of multispectral satellite and drone images was done. The experimental results show that the high-resolution drone imagery provides better crop area estimation and vegetation indices compared to freely available coarse resolution satellite imagery due to mixed pixels especially boundary of the crop classes.

Early-stage wheat detection is quite useful and important for monitoring national food security and crop management systems. However, it is difficult to differentiate early-stage wheat among the ploughed and bare lands using medium to low-resolution satellite data at a large scale. Further, the estimated wheat sowing area would be erroneous due to misclassifications of one class to the other class. On the other hand, drones provide high-resolution data at the field level. Hence, there is a need to fuse drone and satellite data for precision agriculture information at a large scale. Therefore, in this paper, we have explored the application of machine-learning classifiers for early-stage wheat detection using drone and sentinel 1A SAR data. Further, the area is estimated for each class and is compared with the area estimated from the high-resolution drone image. Both qualitative and quantitative analysis of the obtained classification results is carried out with the help of field survey data.

Classification of synthetic aperture radar data has been attempted by researchers in the past for several years, with varying degrees of success. One of the most challenging problem in such classifications is the segregation of urban (built-up region) and tall vegetation (TV). This is due to lack of difference in their scattering power. Therefore, an attempt has been made to improve urban and TV classification accuracy by exploring the use of Sentinel-1 interferometric coherence images as an additional feature along with other popularly used scattering features. A convolutional neural network (CNN) is trained on these features to obtain a good accuracy for classifying urban, TV and other classes. Obtained results are analyzed in both visual and quantitative ways. It is observed that when interferometric coherence images are added as an additional input feature in the CNN, producer and user accuracy for the urban and TV are improved from 15 to 25%, and overall accuracy(OA) is reached from 72.5% to 79.3%. Results oabtained with the CNN model is also compared with the traditional machine learning models such as support vector machine(SVM) and k-nearest neighbors(KNN). CNN over performes with 79.3% OA, followed by KNN with 66% OA while the performance of SVM is the worst among them all with 60% OA.

With the advancements in technology made in the 21 st century, object detection has attracted a lot of attention in recent years. It is probably the most well-known term within the domain of computer vision and it encounters some really interesting problems. In order to gain complete image understanding, we classify different images along with also trying precise estimation of the objects and the locations of these objects contained in each of the images. Object detection is a collection of these related tasks for identifying objects in digital photographs. In this work, we perform detection of different track assets in drone images. For this purpose we have utilized a pretrained model as they are convenient for serving the purpose. The classes in which the images can be categorized into are listed as ‘Construction’, ‘Power Junction/Brick’, ‘Cement Slabs’, ‘Transformer wires’, ‘Garbage’, ‘Person’. We have used “YOLO”v3 (You Only Look Once) for track asset detection. We have initially given an image as input to YOLOv3 and the framework then divides the input image into grids. Image classification and localization are then applied onto each grid. When an image is provided as an input, YOLOv3 framework predicts the bounding boxes and their corresponding class probabilities for objects (if any are found).

Instance segmentation of sleepers is essential in vision-based railway track monitoring to ensure good condition and enumeration. Sleepers for railway tracks are critical as they serve as the foundation for all other auxiliary components. Segmentation of the sleepers in UAV images is challenging due to the complex environment of railway track, occlusion by rail and ballast and viewpoint variation in UAV images. Region-based convolutional neural networks (RCNNs) are popular recently to segment the distinct objects in images based on mask generation methods. This paper applies an efficient approach to segment the rail track sleeper based on Mask-Rcnn with backbone ResNet101 + Fpn, which acts as a feature extractor in images captured from UAV. The results show that sleeper's efficient segmentation and this approach are adaptable for the vision-based track monitoring systems using UAV in real scenarios.

Curating a precise decision-based classifier algorithm to automate target detection based on feature extraction(s) in UAV imagery can assist in various scientific and practical applications. Localization and detection of weed in sugarcane field is a critical classification problem. Vegetative stage of weed, especially, when it is growing and is at its earliest phase, exhibits challenging characteristics such as small weed patch area and color merging tendencies with the crop, which makes it a very typical task to correctly identify, localize and detect weed. A meticulous and scientific detection of weed at early stages may aid in providing timely and quick treatment in the scene to preserve crop health. Random forest classifier is a combination of numerous decision tree classifiers and is a type of ensemble learning which has ample potential for clustering data of similar nature into different classes. This predictive averaging approach has the capability to detect early weed patches, which in turn facilitates precision agriculture. The presented research focusses on binary classification of UAV data of weed infested sugarcane field using decision based random forest classifier at weed's premature stage. Small and multiple green on green weed patches in sugarcane field have been accurately detected and classified into two classes “weed” and “crop”. This algorithm helps detect early weed patches in agricultural setting which in turn aids in weed removal strategies.

Intra field classification of agriculture fields is challenging using dual pol sentinel 1 data due to its limited spatial information. With the use of sensors mountable on unmanned aerial vehicles (UAV) that has a fine spatial resolution, it is easy to obtain intra-field information. However, the use of drones is expensive and their coverage is limited. Therefore, an information fusion approach is used to make intelligent use of high-resolution drone and freely available Sentine1-1 data to provide intra-field classification at large scale for precision agriculture monitoring. Supervised random forest, support vector machine and k-nearest neighborhood classifiers were trained using the polarimetric parameters obtained from dual pol Sentine1-1 data for the classification of sparse and dense vegetation in a large sugarcane field. The overall accuracy of the proposed method is above 84% in segregating sparse and dense sugarcane fields.

Convolutional neural networks (CNNs) are top-rated to classify hyperspectral images. Usually, these use the spectral-spatial approach (SSA), in which the patch corresponding to each pixel to be classified extracted from the hyperspectral image. The size of the patches' spatial neighborhood plays a vital role in the complexity of the designed CNN model. The complexity of the model proportionately increases according to the spatial size of patches. Generally, patches are of odd-squared spatial size centering at the corresponding pixel. In this paper, a novel approach based on mirror mosaicking (MMA) has been proposed. It has been compared with the spectral-spatial approach using minimally sized patches. The proposed approach has been proved computationally efficient along with competitive classification performance. A dataset provided by National Ecological Observatory Network (NEON) has been used for the experimentation, which has three major classes, viz. vegetation, soil, and road.

The accuracy of estimated fractional vegetation cover (FVC) depends on selecting the best suitable input features, precise ground information, and a prediction model. Therefore, in this paper, four machine learning (ML) algorithms, namely Support Vector Regression (SVR), Random Forest Regression (RFR), K-Nearest Neighbors (KNN), and Linear Regression (LR), are used for FVC estimation using different vegetation indices (VI) as input features. Estimated FVC is compared with the ground truth FVC, which has been calculated with the high-resolution drone images, and their R-square values are calculated. R-square value is used for the assessment of the best input features and model. RFR and KNN emerge as the best suitable ML models in comparison to SVR and LR. The obtained R-square values for the RFR model with the input features used as NDVI, PAVI, SAVI, and MSAVI are 0.873, 0.869, 0.869, and 0.862, respectively. NDVI, PAVI, SAVI, and MSAVI perform better in comparison to other features. Hence, they are permuted together and used as input features, and their results on all the algorithms are moderately better. The highest R-square value obtained is 0.878 when SAVI and PAVI are used as input features for the RFR model.

Railway track sleepers are the most critical component that serves as the backbone for other supplementary components. Frequent monitoring of sleepers ensures better railway track health conditions that ensure the safety of goods and passengers. Recently railway explored various monitoring possibilities based on UAV for the robust, cost effective, and efficient inspection of railway track components. Deep learning based object detectors show promising results on large and small datasets due to increased computation resources. YOLO versions of object detection model are most superior on low altitude aerial images. This paper explores the possibility of railway track sleeper detection using a custom object detection model based on the YOLO v4 algorithm in low altitude UAV images.

The detection of faults in any metallic structure in the industries is of great interest. Robust fault detection in different scenarios is necessary to predict the life of the structure. Penetration of the wave is possible through the rusted area at a frequency range 30-300 GHz i.e. at millimeter wave frequency range. Non-Destructive Testing is very useful in industry since it provides the evaluation of the material without causing damage to the original part. This paper is about the millimeter wave image based fault detection in the metallic structure. Conventional fault detection techniques may give a false alarm for the intensities below the average values. A novel approach based on line enhancement filtering is introduced and developed images are compared with other image based processing techniques. Our proposed method gives satisfactory results for a fault like crack having different dimensions in the metal sheet.

The estimation of crop area in advance takes us a step closer towards the intelligent farming as it is beneficial in both pre and post harvesting scenarios for better utilization of resources and higher production at a reasonable cost. There are many challenges in the estimation of crop area in freely available low resolution satellite images due to mixed pixels, especially in the boundaries of crop classes. A neural network has the ability to learn from unknown patterns in the satellite images and then take a decision based on their learning. Fuzzy logic is used together with a neural network that can explain partial membership of each class. Hence, in this paper, we integrate these two models and found it useful to perform accurate estimation of area for each crop class. A quantitative analysis is performed with the help of reference data created by drone images and global positioning system field survey. This study indicates that the proposed method improves the accuracy of area estimation for the crop classes.

Classification of mixed classes that are having similar backscatter response at different polarization combinations, using single frequency synthetic aperture radar (SAR) data is very intricate and there is always a high possibility of misclassification. Therefore, the main objective of this study is to classify the mixed classes using multi-frequency SAR data. An artificial neural network (ANN) approach is used for classification of the considered mixed classes using various polarimetric parameters obtained from single acquisition ALOS2 PALSAR (L band) and Sentinel 1 (C band) dual pol SAR data. An image statistical measure based separability index analysis is used to identify the optimal polarimetric parameters for developing the classifier. It is observed that, the proposed multi-frequency approach is able to classify the mixed classes with an overall accuracy of 87%.

Soil moisture(SM) estimation in agricultural area using microwave data is still a challenging task; due to numerous sources of signal contributors present other than soil dielectric like vegetation, soil roughness, etc. For minimization of the effect of these sources, there is a requirement for fully polarized SAR data or multi-sensor data, i.e., optical and microwave. But the optical data is not easily available in cloudy season, and fully polarized SAR data is less available in comparison to dual-polarized data like Sentinel-1, which is available with a high revisit cycle. Therefore in this paper, an attempt is made to use Sentinel-1 data to estimate SM by minimizing the effect of other signal contributors. Also, for the minimization of SM error, a complete wheat crop cycle is considered, and by using constrained nonlinear multivariable optimization technique, optimum values of WCM parameters and soil roughness are estimated. Obtained results are compared with ground truth SM. Results show that the retrieved SM is very close to ground truth SM.

High precision remote analysis of biochemical variables can enable scientists to obtain very crucial information about crop plants. This information can help in the development of future strategies to protect the crops from different biotic and abiotic stresses thus improving the growth, yield and productivity of crops. Using Sentinel-2 data in combination of drone data for assessing biochemical variables in crop plants is the best choice for timely and accurate monitoring of these variables in the field. As Sentinel-2 data have high temporal as well as spatial resolution and also visible, infrared and red-edge region spectral bands which are highly sensitive to different vegetation properties. It is observed that these bands not only solve the problem of assessing the different crop variables, but its derivatives can provide detailed information on spectral features of crop plants in the form of vegetation indices. Also, for assessing the crop biochemical variable with satellite imagery, the challenge is the accurate identification of that particular agriculture field where the crop has to be monitored. For this purpose, precise agriculture field information is required, which can be obtained using drone images. Therefore, in this paper, a study is conducted to explore the possibility of Sentinel-2 band derivatives, i.e., vegetation indices in analyzing the sensitivity for biochemical variables of different varieties of sugarcane crops. A correlation analysis is carried out for sensitivity analysis between satellite-derived parameters and crop plant biochemical variables. It is observed that the maximum biochemical variables were quite sensitive with different vegetation indices.

Joggled fishplate is an essential component that helps with emergency repairs to fractured sections of the rail. The monitoring of this component is essential in order to maintain steadiness of the train. Manual operations involve railroad personnel. These operations are subjective and time-consuming. Consequently, replacement of manual inspections with unmanned vehicles such as drones is well suited for fast and efficient maintenance operations. Therefore, in this work, we have proposed a critical analysis for development of a novel feature-based template matching technique for efficient and fast joggled fishplate detection in drone data. The appropriate feature is chosen for performing sliding window-based template matching using Euclidean distance. Non-maximum suppression has been used for reduction of false detections. The specificity and sensitivity values are calculated as 0.98 and 0.79 respectively.

In recent years, reliable rail track health monitoring and localization requires accurately orthorectified and georeferenced imagery. The vision-based approach is most suited for the geometrical correction of unmanned aerial vehicles (UAVs) high-resolution imagery in the given scenario. The single image acquired by UAV covers a significant area and contains only one reference point and many distorted pixels. This paper provides a novel computational vision-based approach for orthorectification and georeferencing of a single rail track aerial image among the set of given images without an exclusive reference map of that location and ground control points.

Satellite imagery has been used in most of the applications to estimate, monitor and predict various situations and hurdles in the current scenario. According to the application, the satellite data is chosen on the basis of its spatial, spectral and temporal characteristics. From agriculture to the road highways construction and their health monitoring, it is playing a major role by providing its features details from a high level to medium and low level. In the context of agriculture, there are many ways to find crop health like visual interpretation, health parameter, existing vegetation indices, etc. For the purpose of monitoring and predicting, we require band values from the satellite data which may lack in providing accurate values because of the several parameters that affect it by which there is a possibility of error because they have to be calibrated. Therefore in this paper, a machine learning based calibration approach is proposed by considering multispectral drone data as in-situ data. The methodology is developed and tested for Sentinel-2 data. It is observed that the proposed calibration technique has good potential to calibrate satellite data.

Subclasses classification is one of the major challenges in remote sensing (RS) scene classification. The area under observation, in order to classify agriculture and urban subclasses, requires efficient classification algorithms. Among such algorithms, deep learning algorithm based on Convolutional Neural Network (CNN) architecture is one such promising candidate to obtain the classified map. In this work, performance of a CNN network has been demonstrated on the data obtained from National Ecological Observatory Network (NEON) field site Domain 17 by considering different modality data and its subsequent fusion using the proposed model of CNN as applied on (i) the Hyperspectral, (ii) the Light Detection and Ranging (LiDAR) and then (iii) fused data respectively. Both the Hyperspectral and the LiDAR data have been fused at pixel level. Using the proposed methodology, a classified map is obtained with an overall accuracy of 96 percent for fused data.

Land cover classification in the remote sensing has been done using various deep learning algorithms; and higher classification accuracies have been achieved. Such classification is based on the maximum membership fraction (MMF), when we use Convolutional Neural Network (CNN). MMF is basically the maximum probability fraction. A pixel under prediction has been assigned to that class which has maximum fraction out of the corresponding fractions for all land cover classes. Various methodologies exist for pure pixel extraction and used for hyperspectral unmixing. An assumption has been taken that MMF and abundance used in the case of unmixing are similar. Both MMF and abundance follow the rule of sum to one. In this paper, a classification method has been implemented using CNN to achieve better classification accuracy. Thereafter number of pure pixels extracted based on the various MMF thresholds.

In the last few years' digital agriculture has bloomed at new height. In the digital agriculture formers receive information of different parameters related to their crops in their mobile, and take steps accordingly. The knowledge of the soil moisture (SM) is one of the key parameter related to the agriculture field, and therefore SM is a key parameter for digital agriculture. In the agriculture field, SM varies very frequently, so for digital agriculture, near real time SM needs to be closely observed. For this purpose, microwave sensors which have very high temporal resolution are needed. Recently launched Scatsat-1 Scatterometer, by ISRO, may be an apt data source because of its daily acquisition. However it has some limitations due to its spatial resolution. Therefore, in this paper an attempt has been made to develop such a methodology by which one can retrieve the SM with coarse resolution data. To overcome coarse resolution issue, a methodology is proposed, which considers the vegetation fraction cover (FVC) in every resolution cell. With the help of FVC, backscattering signal of soil (σsoil) is segregated from the total backscattered signal and, this σsoilis inverted to SM using Dubois model. Retrieved SM is compared with ground measured SM, and obtained RMSE is 0.105 m3/m3. Retrieved SM is also compared with available SMAP SM product and observed that retrieved SM is very close to SMAP SM.

Satellite imagery is increasingly being used for the important application of digital agriculture. Efficient estimation of the areas of various land covers such as water, urban, wetland, bare soil, short vegetation and tall vegetation is needed for proper planning of various agriculture activities such as availability of resources, estimating cultivable area and grain yield etc. on a large scale in an unsupervised manner. It is further desirable that the developed techniques should rely on observations acquired using only one sensor that provides data 24x7x365 circumventing the problems of procuring data from multiple agencies, co-registering multiple sensor data etc. Hence, in this paper, three commonly used pixel based fusion techniques spatial frequency SF), principal component analysis PCA and expectation maximization EM algorithm are critically analyzed for fusing multi-polarized high resolution PALSAR data channels among themselves for the agriculture applications. K-means unsupervised algorithm has been applied for classifying various land covers. Unsupervised classification approach as an important step towards development of an autonomous land cover information system with minimum human intervention.

Crop classification is an important information for agriculture monitoring. Crop classification using synthetic aperture radar (SAR) has been performed in many studies, but for agricultural areas like in India that are very fragmented, it becomes difficult to distinguish between two crop types that lie next to each other. Therefore, in this paper, an attempt has been made to classify sugarcane and paddy crops based on object based crop phenology approach for which object based segmentation is explored. The proposed methodology is developed and tested on Sentinel-1 and Sentinel-2 data. It is observed that the proposed methodology has a potential to classify the crop types with an accuracy of 82%.

Normalized Euclidean distance (NED) and normalized signature correlation mapper (NSCM) are most popularly used pattern classifiers with polarization signatures (PSs) based polarimetric synthetic aperture radar (PolSAR) data applications. These methods are not able to fully exploit the PSs as they do not exploit the spatial context or pattern of PSs which is essential. Improved utilization of PSs is still required for PolSAR applications such as agriculture crop classification and monitoring. In this study, convolutional deep neural networks (C-DNNs) are introduced and utilized as pattern classifiers for PS classification. C-DNNs have the ability to consider and control the influence of local neighborhood pixels during classification. Therefore, in this study C-DNNs are utilized to extract and exploit subtle changes between PSs of land covers to improve classification performance. Comparison with NED and NSCM classifiers signify the contribution of C-DNNs by improved performance in PolSAR data classification.

Fractional vegetation cover (FVC) is usually referred to as an important parameter for vegetation health monitoring and also used as control parameter in terrestrial ecosystem change detection. In recent year several models have been developed for FVC measurement using satellite data and digital images at regional and global scale. For the validation and modification in these models need to a precise ground truth information. FVC measured using digital camera act as an efficient ground truth information, but it is also lack in accuracy due to limited number of images and sampling points are possible to take with camera. Drone is the recent trend for precision agriculture monitoring and can be used as substitute to overcome these problems. In this paper an efficient method of ground truth FVC measurement using drone image is developed, which is further used for development of a sigmoid model to measure FVC using Sentinel-2 data for larger area. Results of the obtained model are compared with ground truth FVC and obtained value of RMSE is 0.10. FVC are also measured with dimidiate pixel model and obtained RMSE value with ground truth FVC is 0.17. Results show that developed model can be used for efficient measurement of fractional vegetation cover.

With the recent launch of Sentinel-1 constellations and frequent availability of C-band synthetic aperture radar (SAR) data at no cost, there is an opportunity to monitor crops on regular basis, which is still not explored. Therefore, in this paper an approach has been proposed to monitor sugarcane harvest status using time series Sentinel-1 data. The proposed approach uses knowledge based classification and temporal profile for obtaining harvest status of crops. The approach is able to identify harvested and non-harvested sugarcane areas from other crops with an overall accuracy of 82.17%.

There have been several investigation about water-ice depositions on the lunar polar regions. Earlier, studies were based on criterion circular polarization ratio (CPR). However, It is quite challenging to classify water-ice deposits on the basis of criterion , because it occurs in water-ice and rough surface region both. Thus, it is essential to examine the , laterally with other significant parameters fractal dimension and conformity coefficient (μ) for better classification. First fractal dimension (D) based method has been used to differentiate between the rough and smooth surface. Further, conformity coefficient (μ) is used to identify possible water-ice region associated with volume scattering. This dominant volume scattering pixels points were extracted from the degree of polarization (DOP) for better classification. Finally, obtained results have been compared with existing methods. The entire study indicates that the classification of water-ice deposits using conformity coefficient (μ) gives good results.

A core requirement of database engine testing is the ability to create synthetic versions of the customer's data warehouse at the vendor site. Prior work on synthetic data regeneration suffers from critical limitations with regard to (a) scaling to large data volumes, (b) handling complex query workloads, and (c) producing data on demand. In this demo, we present HYDRA , a workload-dependent dynamic data regenerator, that materially addresses these limitations. It introduces the concept of dynamic regeneration by constructing a minuscule memory-resident database summary that can on-the-fly regenerate databases of arbitrary size during query execution. Further, since the data is generated in memory, the velocity of generation can be closely regulated. Finally, to complement dynamic regeneration, Hydra also ensures that the process of summary construction is data-scale-free.

Fully polarimetric synthetic aperture radar based land cover classification has been intensively investigated for past several decades, but it is still a challenging task to segregate tall vegetation and urban because scattering mechanism involved for both the classes is not sufficient to get the proper threshold in order to differentiate them. Therefore, there is a need to develop such a technique that has the capability to classify these classes with significantly better accuracy. Textural information of an image is known to be an alternate source of extracting useful information of targets. While dealing with natural targets, such as tall vegetation, characteristic of textural feature, i.e., roughness, may be an important parameter which could identify these targets, since both the classes possess different types of roughness. Henceforth, commonly used texture features, i.e., fractal dimension, lacunarity, Moran's I, entropy, and correlation were critically analyzed and realized that these features are still lacking in the concerned segregation, because generally they are pixel-based. Consequently, neighboring pixels are taken into account and an approach has been developed by considering the randomness response (or manner of distribution of scatterers) based on relative similarity of total backscattering power of neighboring pixels by proposing a similarity entropy feature. An optimized threshold method is also developed by means of the contextual thresholding in order to provide a proper decision boundary between the two classes. The proposed approach is successfully tested and validated on different Phased Array type L-band Synthetic Aperture Radar data with sensitivity of tall vegetation and urban as 0.93 and 0.932, respectively.

Hyperspectral images have recently become one of the finest basis for highly accurate identification of objects. Such images, however, are very large in size and carry huge information. Processing and handling of such information is also quite resource savvy and require complex algorithms as well. In this paper, an intelligent classification method has been proposed. Using two independent, simple feed-forward artificial neural networks (ANN), a 426 band hyperspectral image has been first processed for dimensionality reduction to 15 principal components (PCs) using standardized principal component analysis (SPCA), containing 98.86% of the original information. In the second stage, these 15 PCs have been utilized to train another ANN and three different objects viz. road, soil and vegetation have been accurately identified using supervised learning. Only a portion of hyperspectral image data was used as training set and three unique signature patterns were created in the form of pattern library for road, soil and vegetation. Once trained, three independent hyperspectral images taken from San Joachim field site located in California (NEON Domain 17) were fed into the well trained two-stage ANN. The three aforesaid objects were correctly identified. The proposed method is fully scalable and a pattern library can be created to identify more classes of objects. Once trained, the proposed method does not require any more statistical process and all subsequent images can be processed on-board due to the method's implementation using ANNs.

SAR data is playing key role in monitoring, the current status or change in, the land cover. For unsupervised SAR image classification, polarization signatures can play a significant role. Since it is difficult to obtain specific polarization signature of real land cover, it is customary to represent them with standard canonical structures polarization signatures. A critical analysis of the complex signatures of real targets is essential thereafter it is also a challenge to decide the thresholds or class boundary value on the correlation images. Therefore, in this paper an attempt has been made to critically analyze the polarimetric signature of complex targets and based on the correlation image analysis an OTSU multi-thresholding based approach is proposed to decide the individual class boundary values which will finally help in building a decision tree (DT) based classification technique. For this purpose L band fully polarimetric SAR data (PALSAR-2) has been used. DT class thresholds are computed using OTSU multi-thresholding method, scatter plot method, and a priori information. Obtained results reveal that complementary features like polarization signatures can help in identification as well as classification of land surface objects significantly by the proposed method.

Researchers are attempting to classify the fully polarimetric synthetic aperture radar data utilizing diverse methods based on polarimetric indices, decomposition, and image analysis. Although, all these techniques have great potential, and able to segregate distinct land cover classes, but still there occurs ambiguity in classifying urban and tall vegetation classes as they both show similar kind of double-bounce scattering characteristics. In the past, Yamaguchi has modified the polarization process by applying the deorientation effect, which has enhanced the double-bounce characteristic of similar classes like urban and tall vegetation. Based on this, an attempt has been made in this letter to remove the uncertainty between the two classes by proposing a deorientation feature-based classification algorithm which could segregate urban and tall vegetation in an unsupervised way. In this, along with polarimetric, other features, namely, color, texture, and wavelets, have been critically analyzed on the deoriented image as each feature type has its own points of interest and hindrances. The result obtained from the proposed technique has shown good accuracy rate for urban and tall vegetation classification.

This paper critically analyzes several incoherent model-based decomposition methods for assessing the effect of deorientation in characterization of various land covers. It has been found that even after performing decomposition, ambiguity still occurs in scattering response from various land covers, such as urban and vegetation. Researchers introduced the concept of deorientation to remove this ambiguity. Therefore, in this paper, a critical analysis has been carried out using seven different three- and four-component decomposition methods with and without deorientation and two Eigen decomposition-based methods to investigate the scattering response on various land covers, such as urban, vegetation, bare soil, and water. The comprehensive evaluation of decomposition and deorientation effect has been performed by both visual and quantitative analyses. Two types of quantitative analysis have been performed; first, by observing percentage of scattering power and second, by analyzing the variation in the number of pixels in different land covers for each scattering contribution. The analysis shows that deorientation increases not only the power but also the number of pixels for surface and double bounce scattering. The number of pixels representing volume scattering remain almost the same for all the methods with or without deorientation, whereas volume scattering power reduces after deorientation. Eigen decomposition-based methods are observed to solve the problem of overestimation of volume scattering power.

This paper presents the feed-forward back-propagation technique of artificial neural networks to validate the false alarm detection in breast cancer. A simple model of the human breast with a tumor of size 5 mm has been used to record the data of electromagnetic wave scattering in the microwave band in range 1-10 GHz. The recorded data with and without added synthetic noise is used to train, validate and test the networks. The synthetic noise is added to the data to validate the ruggedness of the ANN model. The increase of noise in recorded signal with deterioration of the signal to noise ratio from 40 dB to 1 dB in signal results the increase in false alarm detection from 2% to 22%, which can become the cause of the false report generation of the breast cancer detection.

Presently, there is a need to explore the possibility to maximize the use of MODIS (Moderate Resolution Imaging Spectroradiometer) data as it has very good spectral (36 bands) and temporal resolution whereas its spatial resolution is moderate i.e. 250m, 500m, and 1km. Because of its moderate spatial resolution, its application for land cover classification is limited. Therefore, in this paper, an attempt has been made to enhance its spatial resolution and utilize the information contained in the different bands together to achieve good land cover classification accuracy, so that, in future, MODIS data can be used more effectively. For resolution enhancement, modified dual tree complex wavelet transform (DT-CWT) has been employed, where DT-CWT has been modified by critically analyzing the effect of weight factor of the DT-CWT coefficients on land cover classification. For this purpose, image statistics parameter like Mean of the image has also been considered. The proposed technique has been applied on the six bands of MODIS data which have spatial resolution of 500m. It is observed that weight factor of the high-frequency sub-bands is quite sensitive for computation of classification accuracy.

Conventional methods for classifying SAR data, such as H-α decomposition, Wishart classifier etc. are quite complex and classifies data only on the basis of polarimetric information. With the advent of distinct feature types, their role in land cover classification using SAR data could be analysed. For the sake of classification, researchers are extracting and combining several features in order to obtain the best attainable accuracy. But the usage of several feature type is not only increasing the computational complexity, but also the salience of each of the feature type remains unhighlighted. Hence, it became difficult to analyse that which feature type are best suitable for classification and selection of suitable features for land cover classification is challenging as each feature has its own significance level. Therefore, in this paper class wise, optimal feature selection for land cover classification has been performed using SAR data. For optimal feature selection, four types of feature set polarimetric features, texture features, color features and wavelet features have been examined. For class wise feature subset selection separability index criteria and classification results obtained using Naive Bayes classifier has been utilized. With the proposed methodology overall 10 features has been selected among the total 37 feature analysed with fine land cover classification accuracy of 91%.

Survival from breast cancer strongly linked to the size of the tumor at the detection stage. Thus, the early stage detection of tumor of size as minimum as 1.0 mm radius is of great research interest. Currently used techniques for breast cancer detection fails in 10-30% cases and it gives any positive results when the tumor grows in to a size more than 10.0 mm, this reduces the possibility for an early stage detection and thus the survival rate. Thus, in this paper an alternate method of breast cancer detection through microwave imaging is studied. A dielectric mixing model is used to compute the dielectric constant of the breast tissue with and without the malignant tissue and the proposed model is verified through the simulation in CST. Free space transmission and metal back method are used for the measurement of dielectric constant of the phantom containing one, two, three and four tumors of radius 1.0 mm each. The proposed dielectric mixing model can be applied to detect the changes in the dielectric constant of the tumor affected tissue of radius 1.0 mm which is not possible through any other existing methods.

The concealed weapon, like blade, detection and identification is one of the most puzzling task faces by security agency. Researchers have demonstrated MMW imaging systems to detect concealed targets like gun, knife and scissors but detection of small size target like blade with different orientation is still challenging due to resolution limitation of MMW imaging system. The success of small size concealed target detection depends upon scanning step size of imaging system and dielectric property of covering cloths and hidden object. Therefore, resolution enhancement techniques may play a very important role for small size concealed target detection. To perceive such challenges, active V-band MMW radar conjunction with image processing techniques has been demonstrated for detection and identification of concealed blade and obtained two dimensional good quality of images of concealed blade under different cloths at various angle. For this purpose, a critical analysis of various signal and image processing has been carried out and integrated following algorithms like singular value decomposition (SVD) for clutter reduction, discrete wavelet transform (DWT) for resolution enhancement, thresholding for target detection and in last artificial neural network (ANN) based algorithm for rotation invariant target identification. An image processing based methodology has been proposed by which the concealed target like blade can be successfully detected.

Currently, mean subtraction, median removal, singular value decomposition (SVD), Principal component analysis (PCA) and Independent component analysis (ICA) areverypopular approaches to extract the buried target information in presence of clutter and background noise for GPR applications. Clutter and background reduction and detection of low dielectric constant buried object with variable soil conditionsare the challenging tasks in GPR. But available techniques are not able to extract the non-metallic target information, due to low dielectric constant. Therefore, this paper proposes a neural network and statistical mean to standard deviation threshold based approach for subtracting background and for enhancing the detection of low dielectric constant buried object. ANN approach is based on the collection of large amount background data with soil moisture variation. These background data statistically analysed to compute the mean to standard deviation thresholding. After that, motion filter estimate the actual pixel intensity of PVC pipe in linear manner. The results show that the enhanced target detection and background subtraction are achieved directly from proposed trained neural network.

In the past SAR data has been proven as a great source for land cover characterization. For classification purpose many individual methods has been used, but single method are likely to undergo high variance or biasness depending on the base used for classification. Hence, in this paper random forest classification technique has been used for SAR data classification into different land cover classes (urban, water, vegetation and bare soil) which minimizes the diversity amongst the fragile classifiers and produce more accurate predictions. In this regard, an attempt has been made to fuse, four types of measures, namely texture features, SAR observable, statistical features and color features using random forest classifier for land cover classification. The results show that the resultant classified image has better accuracy in comparison to the individual method.

The application of Synthetic Aperture Radar (SAR) data for estimating soil moisture in the vegetated covered areas is still challenging task because variation of the vegetation over the soil surface is difficult to predict. Thus, there is a need to develop such an approach for soil moisture retrieval which can minimize the vegetation effects. Various models are available in the literature for soil moisture retrieval but either these models are very complex or they required more than one satellite data. Therefore, in this paper, an attempt has been made for minimizing the crop effect while retrieving the soil moisture by using same satellite data (i.e., PALSAR data) only. For this purpose, the application of polarimetric information may be useful because polarimetric indices give physical significance in discriminating the vegetated cover areas from other regions (water bodies, urban areas, bare soil) of the earth's surface. So, these polarimetric indices have been critically analyzed for the minimization of crop effect and a model has been developed for soil moisture retrieval by using the polarimetric indices SPAN and RVI.

Classification of water ice region on lunar surface with Mini-SAR data is quite challenging. Therefore, a probability density function (pdf) based pattern analysis approach has been applied to classify lunar surface. This paper represents the pattern analysis approach to fit data points to a distribution function for understanding the distribution behaviour of Mini-SAR data which helps in developing a method based on density functions to differentiate two types of craters namely icy (type-I) and non-icy (type-II) craters. Circular polarization ratio (CPR) is a very important parameter in study of lunar surface. More specifically, the criterion CPR>1 is used to determine possible presence of water-ice deposits on lunar surface So, it's important to study distribution behaviour of CPR pixels and to determine best fitted distribution function representing this behaviour. Therefore, in this paper, pattern analysis techniques have been applied to differentiate two crater types based on the distribution behaviour of CPR. The best fitted function for CPR has been obtained as Generalized Extreme Value function which clearly differentiate type-I and type-II craters.

The aim of model based decomposition is to express coherency matrix in terms of various scattering components (like, volume, surface, double bounce, and helix). In spite of this decomposition, ambiguity occurs in scattering response from various land covers, like urban and vegetation. Deorientation process is believed to remove this ambiguity. However, there is a need to check whether decomposition methods and deorientation helps in identification of different land covers in terms of scattering mechanisms. To fulfil this task, in this paper, a study of four D decomposition methods with and without deorientation has been performed. The purpose of this study is to visualize the effect of deorientation on various land covers like, urban, vegetation, bare soil, water, and subsidence, in Jharia region, one of the major coal fields of India. Both visual and quantitative analysis have been performed for comprehensive evaluation of deorientation effect.

The polarimetric information contained in polarimetric synthetic aperture radar (SAR) images represents great potential for characterization of natural and urban surfaces. However, it is still challenging to identify different land cover classes with polarimetric data. Most of the classification algorithms presented earlier have used a fixed value of polarimetric indexes for segregation of a particular land cover type from other classes. However, the value of these polarimetric indexes may change accordingly with change in observation site, temporal acquisition, environmental conditions, and calibration differences among various systems. Thus, the value of polarimetric indexes for segregation of each land cover type has to be tuned in order to cope with these changes. Therefore, in this paper, a decision-tree-based adaptive land cover classification technique has been proposed for labeling of different clusters to their own classes. The proposed method uses spatial-statistics-based expressions (i.e., median “ M” and standard deviation “ S”) of best-selected polarimetric indexes on the basis of a separability index criterion for creating the decision boundary among various classes. In order to make the system adaptive in nature, unknown terms have been included in the expressions. Due to the dependence of a developed nonlinear relationship of overall classification accuracy (OA) on large number of unknowns, a genetic algorithm (GA) approach has been used, which provides optimum values of considered polarimetric indexes for automatic segregation of different classes. The proposed algorithm is successfully tested and validated on ALOS PALSAR quad-pol data.

In the present paper an attempt has been made for estimation of volumetric moisture of bare soil surface at particular thickness of soil layer by applying impedance based model relied on transmission line theory. The proposed approach calculates the impedance of layer of bare soil surface as a function of complex dielectric constant and thickness of soil. After retrieval of complex dielectric constant at 5cm thickness of soil layer using Genetic algorithm (GA) approach, volumetric moisture of bare soil surface is retrieved by using polynomial relationship proposed by Topp et al. The proposed method is effectively applied on radarsat-2 data. Obtained soil moisture values show consistency with the soil moisture values retrieved by Dubois model and ASCAT SSM data.

The present paper deals with the task of identifying lunar craters having possible existence of water-ice deposits on their surface. For this purpose, a decision tree algorithm has been proposed, in which decision criterion are decided on the basis of CPR, m-δ decomposition, fractal dimension `D' and conditions proposed by Thompson et al., The proposed algorithm is successfully applied on Chandrayaan-1's MiniSAR data.

In this paper, we have analyzed the angular response of specular scattering coefficient for different soil texture fields while varying soil moisture and surface roughness at C-band. An approach based on multi-incidence angle data has been developed to retrieve soil texture, soil moisture and surface roughness. An empirical relationship has been developed between normalized specular scattering coefficient and surface roughness parameters. This empirical relationship has been utilized along with the Kirchhoff Scalar Approximation to retrieve soil texture, soil moisture and surface roughness. Obtained results are in good agreement with ground truth data.

This paper presents a novel shape recognition approach for analysis of through wall images. The developed algorithm consists of image formation, processing for image enhancement, feature extraction and shape recognition. In this paper, ultra wideband step frequency continuous wave (SFCW) radar is used to collect C-scan data from which images are obtained to give the approximate shapes of target. Results on experimental data show that the proposed algorithm exhibits promising performances both in terms of target detection and recognition.

In single-band single-polarized SAR images, intensity and texture are the information source available for unsupervised land cover classification. Every textural feature measure identifies texture patterns by different approaches. For efficient land cover classification, textural measures have to be chosen suitably. Therefore, in this letter, the role of various intensity and textural measures is analyzed for their discriminative ability for unsupervised SAR image classification into various land cover types like water, urban, and vegetation areas. To make the algorithm adaptable, these textural features are fused using principal component analysis (PCA), and principal components are used for classification purposes. To highlight the effectiveness of PCA, the difference between PCA- and non-PCA-based classifications is also analyzed. Analysis of the role of texture measures for unsupervised classification of real-world SAR data with application of PCA is presented in this letter. The analysis of how every individual feature measure contributes for classification process is presented, and then, textural measures for a feature set are chosen according to their role in improving classification accuracy. By analysis, it is observed that the feature set comprising mean, variance, wavelet components, semivariogram, lacunarity, and weighted rank fill ratio provides good classification accuracy of up to 90.4% than by using individual textural measures, and this increased accuracy justifies the complexity involved in the process.

Present paper aims to critically analyze and study the SAR interferometric RADARSAT-1 data for land subsidence monitoring system for New Orleans USA, using Differential SAR Interferometry (D-InSAR) approach. In Louisiana, areas along the coast are sinking as much as one inch a year, and New Orleans is among the worst affected area due to land subsidence. For mapping subsidence, 84 complex data sets from April 15, 2002 to March 15, 2007 were available. Out of this, we have chosen six sets of suitable differential interferometric pairs with approximately one year of temporal span and six interferometric pairs to provide DEM generation for the above differential pairs. In this paper, change detection due to surface deformation is identified using D-InSAR methodology and is compared with classical change detection approach using MRD (mean ratio detector). Results show that subsidence was widespread throughout New Orleans, with maximum subsidence near MRGO canal in period of March 01, 2005 to April 1, 2006.

This paper deals with the potential application of National Oceanic and Atmospheric Administration (NOAA)/Advanced Very High Resolution Radiometer (AVHRR) data to detect subsurface fire (subsurface hotspots) by proposing an efficient contextual algorithm. Most of the solutions proposed to date are mainly focused on the problem of surface fires, and very few research works have been performed to develop techniques for the subsurface fire problem. Although few algorithms based on the fixed-thresholding approach have been proposed for subsurface hotspot detection, however, for each application, thresholds have to be specifically tuned to cope with unique environmental conditions. The main objective of this paper is to develop an instrument-independent adaptive method by which direct threshold or multithreshold can be avoided. The proposed contextual algorithm is very helpful to monitor subsurface hotspots with operational satellite data, such as the Jharia region of India, without making any region-specific guess in thresholding. Novelty of the proposed work lies in the fact that once the algorithmic model is developed for the particular region of interest after optimizing the model parameters, there is no need to optimize those parameters again for further satellite images. Hence, the developed model can be used for optimized automated detection and monitoring of subsurface hotspots for future images of the particular region of interest. The algorithm is adaptive in nature and uses vegetation index and different NOAA/AVHRR channel's statistics to detect hotspots in the region of interest. The performance of the algorithm is assessed in terms of sensitivity and specificity and compared with other well-known thresholding techniques such as Otsu's thresholding, entropy-based thresholding, and existing contextual algorithm proposed by Flasse and Ceccato. The proposed algorithm is found to give better hotspot detection accuracy with lesser false alarm rate.

The potentiality of synthetic aperture radar (SAR) images for land cover mapping is an important area of research. For single band, single polarized SAR images, information is available in the form of intensity and texture only. Land cover classification of SAR images requires exploitation of spatial relationship of pixels also, in addition to pixel level segmentation. SAR images can be segmented successfully if the regions with homogeneous intensity and texture areas can be identified and grouped together. So far, contour tracing has been used only in demarcating sea and land. Identifying contours in a domesticated area with a mixture of water, urban and vegetation areas require complex analysis of the spatial distribution of pixels. In this paper, we have presented an unsupervised classification algorithm using maximumaposteriori(MAP) segmentation for SAR images in which SAR image is classified into monotone, texture and edge regions. Monotone and textured regions are labeled as land cover types like water, urban and vegetation areas using K-means classification. SAR image of the region with latitude varying from 77.86deg to 77.91deg and longitude varying between 29.89deg and 29.85deg of Haridwar region, India is considered for segmentation. We have compared the segmented image obtained by this methodology with the topographic map of the corresponding region. The water, urban and vegetation areas are clearly recognized with the proposed classification approach which represents a very good agreement with the original topographic sheet.

In this paper, harmonic analysis of 10-year time series (1995-2005) NOAA/AVHRR yearly composite images is performed to develop an innovative technique for hotspot detection and land-features classification based on temporal changes in the NDVI and various AVHRR band values. NOAA/AVHRR images are used due to wide coverage, high frequency and free acquisition offered by NOAA/AVHRR sensors. Proposed algorithm consists of three steps: (1) preprocessing of NOAA/AVHRR images to correct geometric distortions and calibrate the data radiometrically (2) detection of cloud and water pixels in the preprocessed image and application of harmonic analysis on 10-years time series AVHRR images to produce phase and amplitude images, and (3) application of image processing techniques on the amplitude images of different bands to detect hotspots and classify the region of interest. The obtained results indicate that the proposed method can classify the region of interest successfully with through out greater than 91% classification accuracy.

With the availability of multisensor and multiresolution image data from operational Earth Observation satellites, the fusion of digital image data has become a valuable tool in land cover classification. Digital image fusion is a relatively new research field at the leading edge of available technology. It forms a rapidly developing area of research in land cover classification. It is needed that to fuse high resolution satellite data with low resolution satellite data, to enhance the classification and interpretation in low resolution satellite data. The AVHRR and MODIS data are freely available, but resolution is poor. Therefore in this paper, it is attempted to highlight the AVHRR and MODIS utility with fusion of ASTER data. In this paper, a fusion method based on the Curvelet transform is introduced. The curvelet transform represents edges more accurately, since edges play a fundamental role in image understanding, one good way to enhance spatial resolution is to enhance the edges. Curvelet-based image fusion method provides richer information in the spatial and spectral domains simultaneously.

Present paper deals the fusion of image analysis with electromagnetic and support vector machine (SVM) optimization approach to estimate the depth of shallow buried metallic and dummy mine (i.e., without explosive) objects with microwave remote sensing data at X-band (i.e., 10 GHz). The objects were buried under dry and smooth sand. For this purpose, a monostatic scatterometer at X-band has been indigenously developed, which consists a transmitter and receiver mounted on the stand of the sand pit and when operated it moves over it in X and Y- axis. An algorithm has been proposed for identification of suspected region first i.e., region of interest (RO1) that contains buried objects in the image by proposing a quantity “detection figure” (D), which further proceed for depth estimation of buried objects. Algorithm includes image processing, electromagnetic multi layer interaction and SVM approach. The convolution-using image processing techniques has been applied to avoid the overlapping of the return signal. The support vector machine (SVM) approach has been analyzed for estimation of depth and an efficient method based on electromagnetic multiplayer interaction concept has been proposed to train the SVM. The depth estimated for Al sheet gives better result than dummy landmine, but the estimated depths results for both objects are in good agreement with actual depths. The present approach may be quite helpful to develop an automatic satellite data based information systems to estimate the depth of various shallow buried objects with satellite or air-borne radar data.

An observation has been carried out at different incidence angle by X-band scatterometer for like polarizations (i.e., Horizontal-Horizontal and Vertical-Vertical) to retrieve the soil moisture of bare field. To retrieve the soil moisture with minimizing the surface roughness effect on the soil moisture a polarimetric discrimination ratio (PDR) (PDR = (/spl sigma//spl deg/ VV - /spl sigma//spl deg/ HH)/ (/spl sigma//spl deg/ VV + /spl sigma//spl deg/ HH) where /spl sigma//spl deg/ is scattering coefficient for VVand HH-polarization) approach is proposed. Generally in microwave applications, polarization is sensitive to the size, shape and orientation of target elements, whereas frequency functions primarily as a size filter with important consequences regarding attenuation. In the polarization, the horizontal polarization gives a measure of the horizontal dimension of the emitting elements, while the vertical polarization essentially gives a measure of the vertical dimension. The power observed to the system is a complicated function of each particle size, shape and density, therefore we have normalized the value of HH- and VV-pol to the range -1/spl les/PDR/spl les/1 to partially minimize the effect of orientation and shape on dielectric properties of the scattering elements and results mainly the effect of dielectric (i.e., moisture content) on /spl sigma//spl deg/ . The angular dependence of PDR on moisture is observed and it was found that the suitable incidence angle to observe the soil moisture at X-band with PDR is 35/spl deg/. An empirical relation has been developed between PDR and soil moisture to retrieve the moisture content at X-band. The results are validated with another set of data. The retrieved values of soil moisture are in good agreement (standard error-1.02) with observed values of soil moisture. The significance of the results has been tested by F-ratio test and quite significant results were obtained. This type of results are helpful to use the microwave techniques to assess the soil moisture by the air-borne or space borne sensor and propose the suitable angle of incidence to observe the soil moisture with minimum effect of surface roughness at X-band.

The aim of this study is to assess the bare soil surface roughness parameter, i.e. the root mean square height (h/sub rms/) when the moisture is constant, using anechoic chamber measurements based on fully polarimetric scatterometer data. An incidence angle based algorithm has been proposed to assess the bare soil height h/sub rms/. For this purpose, sets of experimental backscattering data have been evaluated on different types of rough surfaces (Rough Gaussian, h/sub rms/ = 2.5 cm, Smooth Gaussian, h/sub rms/ = 0.4 cm and Medium Mixed surface, h/sub rms/ = 0.9 cm) with known geometrical and dielectric properties. The scattering matrix of those three surfaces under test was measured in monostatic mode vs. frequency ( 1 - 19 GHz) and incidence angles ( /spl theta/ = 10/spl deg/ to 50/spl deg/ in steps of 10/spl deg/ for h/sub rms/ = 2.5 cm and 0.9 cm and 5/spl deg/ for h/sub rms/ = 0.4 cm) data. An empirical relationship has been developed between backscattering, h/sub rms/ and incidence angle independently for L-, C-, X- and Ka-band for all polarizations (i.e. HH, VV and HV). This relationship provides the calculated backscattering values, which is helpful in the inversion process. A good agreement has been obtained between the observed and calculated h/sub rms/. The analyses show the strong dependence of h/sub rms/ on incidence angle, polarization and frequency. This type of work is also helpful in the near future to predict the optimum sensor parameters (i.e. incidence angle, polarization and frequency) for measuring the bare soil roughness.