In the present paper, a new deep neural network model, named, UrbanScape-Net, has been designed for improved land-cover classification. Motivated by the need to emphasize important features and capture intricate relationships in land-use scenes, spatial and self-attention mechanisms with the Xception architecture are integrated. The spatial attention module focuses on salient regions, the self-attention method captures intricate relationships within these regions and the dynamic convolutional layer with varying filter sizes extracts multi scale features. The Xception model, pretrained on ImageNet, has been employed as the foundational feature extractor. The experiment conducted on the UC Merced Land Use dataset demonstrated enhanced accuracy (of 98.83%) across 21 classes compared to state-of-the-art methods.

Hyperspectral sensors obtain a set of images from hundreds of contiguous and narrow bands of electromagnetic spectrum from visible to infrared regions. As large number of bands are present in these images, computational overhead for classifying them is very high. To speed up classification, reducing dimensionality by proper selection of a subset of bands is necessary. A filter based method using mutual information is proposed in this regard. Classification is done using Support Vector Machine classifier. Two evaluation measures, namely, overall classification accuracy and Kappa coefficient are considered to assess the efficiency of the proposed method. Performance of the proposed technique is compared with two other mutual information based methods and the proposed method is found to be better as compared to others.

An extensive collection of algorithms have been proposed over the years to identify the foreground from a video scene, but none of them considers the classification history of previous frames for discovering moving objects. All the existing algorithms focus on a single background model to cope with all types of challenging and complex video environments. In this paper, a real-time pixel level classification method is proposed that uses its previous output history to update its parameters. The model has three components designed to handle various types of challenging background environments. For each pixel, three sub-parts, namely the Adaptive Background Model, the Neighborhood Background Model, and the Change Detection Background Model are constructed to detect various types of complex background changes. A pixel level model updating method uses the previous foreground/background binary classification results to periodically refresh all the three background models. This update method helps to recognize the foreground accurately by adjusting the algorithmic parameters to efficiently detect complex background changes. This novel method shows a significant improvement in performance for a variety of complex video scenes.

Background subtraction (BGS) is a popular scheme epitomized in the state-of-the-art literature on video processing. In this context, a novel online kernelized fuzzy modal variation based background subtraction scheme for detecting local changes from the sequences of image frames is proposed. In the proposed scheme, the time varying background at different instances of time are modeled using fuzzy set theory. The proposed background subtraction scheme, utilizes the fuzzy modal variation as the cost function for fitting the pixel values of the image frames. The use of kernel based modal variation helps in projecting the pixel values in a higher dimensional space, linearly separating them into object and background classes. The results of the proposed technique is verified on different challenging sequences including dynamic background, camera jitter, noise, blurred scene, etc. The proposed technique is successfully tested over several test sequences with two major databases (all sequences) and it provides better results compared to the twenty one existing state-of-the-art techniques.

Latest developments in the field of power-efficient neural interface circuits provide an excellent platform for applications where power consumption is the primary concern. Developing neural networks to achieve pattern recognition on such hardware remains a daunting task owing to substantial computational complexity. We propose and demonstrate a Spiking Neural Network (SNN) with biologically reasonable time constants to implement basic Boolean Logic Gates. The same network can be further applied to more complex problem statements. We employ a frequency spike encoding for data representation in the model, and a simplified and computationally efficient model of a neuron with exponential synapses and Spike Timing Dependent Plasticity (STDP).

Most of the supervised dimensionality reduction (DR) methods design interclass scatter as the separability between the class means, which may force to assume unimodal Gaussian likelihoods and their projection space trends toward the class means. This paper presents a novel DR approach, norm discriminant eigenspace transform (NDET), in which average norms (l2) of classes have been utilized to characterize the interclass separability and the within-class distance characterizes the intraclass compactness. NDET is intended to accommodate data distributions that may be multimodal and non-Gaussian. We derive an upper bound for NDET, and a specific solution space to attain this bound. Existence of the specific solution is very unwonted, thereby we have considered the solution space of upper bound to achieve better reduction of dimensionality and discrimination of classes. Also, a nonlinear version of NDET (kernel NDET) is developed to model nonlinear relationships between the features. We show, experimentally (on synthetic data) that NDET effectively overcomes the limitations, which arise due to unimodal and data distribution assumptions of the traditional algorithms. Extensive empirical studies are made; and the proposed method is compared with closely related state-of-the-art schemes on UCI machine learning repository and face recognition data sets, to establish its novelty.

This paper proposes an efficient real-time background subtraction algorithm, which is essential in many computer vision applications. Initially, histograms of the intensity values for each channel of a pixel position in a set of training frames are constructed. A background model, histogram min-max bucket, is constructed from the minimum and maximum values of contiguous non-zero frequencies of the temporal intensity histogram. A novel feature of this algorithm is the use of a single sliding window to update the system adaptively, capturing the proper background even under sudden and/or gradual illumination changes in the scene. We incorporate both local and global information with the help of the current and the previously visited pixel values for binary classification of a pixel into foreground and background. This algorithm is compared with several state-of-the-art techniques and experimental studies show that the proposed method outperforms all these methods in terms of accurate binary classification.

Dimensionality reduction is an important task where the aim is to reduce the number of features and make the system less time consuming for classification. Here, the drawbacks of Fisher's linear-discriminant-analysis-based feature extraction (FE) methods are addressed and a proposal is made to overcome it as well as to reduce the Hughes phenomenon and computational complexity of the system. The proposed FE technique initially partitions the complete set of features into several highly correlated subgroups. Then a linear transformation is performed using a maximal margin criterion over each subgroup. The proposed method is supervised in nature, because prior information about the class label of data is required to calculate the maximum margin criterion based on interclass and intraclass scatter matrices. Experiments are conducted with the PaviaU and Indian pine data sets, and the results are compared with five state-of-the-art techniques, both qualitatively and quantitatively, to demonstrate the effectiveness of the proposed method.

Though various image segmentation techniques have been developed, it is still a very challenging task to design a robust and efficient algorithm to segment (noisy, blurred or even discontinuous edged) images having high intensity inhomogeneity or non-homogeneity. In this article, a robust fuzzy energy based active contour, using both global and local information, is proposed to detect objects in a given image based on curve evolution. The local energy is generated by considering both local spatial and gray level/color information. The proposed model can better deal with images having high intensity inhomogeneity or non-homogeneity, noise and blurred boundary or discontinuous edges by incorporating local energy term in the proposed active contour energy function. The global energy term is used to avoid unsatisfactory results due to bad initialization. We show a realization of the proposed method and demonstrate its performance (both qualitatively and quantitatively) with respect to state-of-the-art techniques on several images having such kind of artifacts. Analysis of results concludes that the proposed method can detect objects from given images in a better way than the existing ones.

Participation of class-wise noisy patterns may mislead the selection process of relevant patterns for subspace projection. And modelling between-class scatter for each class using the patterns that are nearer to the corresponding class decision boundary may improve the quality of feature generation. In this manuscript, a novel dimensionality reduction method, named Maximum Class Boundary Criterion (MCBC) is proposed. MCBC increases class separability by realizing the significant class-boundary and class-non-boundary patterns after the elimination of noisy patterns. The objective of MCBC is modeled such that the class-boundary patterns are pushed away from the corresponding class means and class-non-boundary patterns are forced towards their class means. As a result, the classification performance of the extracted MCBC features is improved. Experimental study is performed on UCI machine learning and face recognition data to highlight the performance of MCBC. The results conclude that MCBC can generate better discriminative features compared to the state-of-the-art dimensionality reduction methods.

The aim of this letter is twofold. First, we assess the compressive sensing (CS) approach as a classification tool for multispectral remote sensing images, assuming severe scarcity of training samples (at most, ten for each class). Then, we propose a new strategy to perform domain adaptation using a CS approach for classifying images at large spatial scales (continental mapping). In particular, the “most confusing” training samples in the target domain are collected by exploiting plenty of training samples available in the source domain under the transfer learning framework. For assessing the proposed method, experiments are performed on three remotely sensed images captured by the Landsat 8 satellite in different regions of India. Results obtained using the proposed approach are found to be promising.

In this letter, a change detection technique using a multiple classifier system is proposed. Here, different architectures of multilayer perceptron (MLP) are used as base classifiers. An ensemble of different MLPs is utilized to increase the robustness of the system. This also avoids the problem of choosing an optimum architecture for MLP. First, the support values for each of the unlabeled patterns are estimated using different MLPs (trained with the labeled patterns). Then, each of the unlabeled patterns is assigned to a specific class by fusing the outcome of the base classifiers using different combination rules. Experiments are carried out on multitemporal and multispectral images. Results show that the proposed ensemble technique has an edge over individual base classifiers for change detection in remotely sensed images.

In this article a spatio-contextual unsupervised change detection technique for multitemporal, multi spectral remote sensing images is proposed. The technique uses a Gibbs Markov Random Field (GMRF) to model the spatial regularity between the neighboring pixels of the multitemporal difference image. The difference image is generated by change vector analysis (CVA) applied to images acquired on the same area at different times. The change detection problem is solved using the Maximum a posteriori probability (MAP) estimation principle. The MAP estimator of the GMRF used to model the difference image is exponential in nature, thus a modified Hopfield type neural network is exploited for estimating the MAP. In the considered Hopfield type network, a single neuron is assigned to each pixel of the difference image and is assumed to be connected only to its neighbors. Initial values of the neurons are set by histogram thresholding. An Expectation Maximization (EM) algorithm is used to estimate the GMRF model parameters. The proposed technique is validated by testing on different multispectral and multitemporal remote sensing images and compared with existing state-of-the-art techniques.

Histogram equalisation has been a much sought‐after technique for improving the contrast of an image, which however leads to an over enhancement of the image, giving it an unnatural and degraded appearance. In this framework, a generalised contrast enhancement algorithm is proposed which is independent of parameter setting for a given dynamic range of the input image. The algorithm uses the modified histogram for spatial transformation on grey scale to render a better quality image irrespective of the image type. Added to this, two variants of the proposed methodology are presented, one of which preserves the brightness of original image while the other variant increases the image brightness adaptively, giving it a better look. Qualitative and quantitative assessments like degree of entropy un‐preservation, edge‐based contrast measure and structure similarity index measures are then applied to the 500 image data set for comparing the proposed algorithm with several existing state‐of‐the‐art algorithms. Experimental results show that the proposed algorithm produces better or comparable enhanced images than several algorithms.

In this paper, a spatiocontextual unsupervised change detection technique for multitemporal, multispectral remote sensing images is proposed. The technique uses a Gibbs Markov random field (GMRF) to model the spatial regularity between the neighboring pixels of the multitemporal difference image. The difference image is generated by change vector analysis applied to images acquired on the same geographical area at different times. The change detection problem is solved using the maximum a posteriori probability (MAP) estimation principle. The MAP estimator of the GMRF used to model the difference image is exponential in nature, thus a modified Hopfield type neural network (HTNN) is exploited for estimating the MAP. In the considered Hopfield type network, a single neuron is assigned to each pixel of the difference image and is assumed to be connected only to its neighbors. Initial values of the neurons are set by histogram thresholding. An expectation-maximization algorithm is used to estimate the GMRF model parameters. Experiments are carried out on three-multispectral and multitemporal remote sensing images. Results of the proposed change detection scheme are compared with those of the manual-trial-and-error technique, automatic change detection scheme based on GMRF model and iterated conditional mode algorithm, a context sensitive change detection scheme based on HTNN, the GMRF model, and a graph-cut algorithm. A comparison points out that the proposed method provides more accurate change detection maps than other methods.

In this article, we propose a novel object detection technique that can separate the moving object from its shadow. In this regard, we initially built a background model by taking median of pixel values in the temporal direction. To suppress the effects of quick change in illumination, and color frequency variation of the textured background, we have extracted the RGB color and ten local features at each pixel location in the target image and background model. For background separation, a difference image is generated by considering pixel by pixel absolute difference of the thirteen dimensional target image frame and the constructed background model. This is followed by a spatial Markov Random Field (MRF) constrained fuzzy clustering to find the moving regions in the target frame. The maximum a'posteriori probability (MAP) of the MRF constrained fuzzy clustering provides a binary image, where the moving objects with the moving cast shadow are identified as one group and the background is obtained as another group. To segment the moving objects from its shadow we explore a three stage shadow analysis technique. It uses analysis of rg color chrominance property of shadow, local gray level feature based shadow processing followed by boundary refinement to separate out the moving objects from its shadows. The performance of the proposed scheme is evaluated by comparing it with the state-of-the-art techniques.

In this paper, we put forward a novel region matching-based motion estimation scheme to detect objects with accurate boundaries from videos captured by moving camera. Here, a fuzzy edge incorporated Markov random field (MRF) model is considered for spatial segmentation. The algorithm is able to identify even the blurred boundaries of objects in a scene. Expectation Maximization algorithm is used to estimate the MRF model parameters. To reduce the complexity of searching, a new scheme is proposed to get a rough idea of maximum possible shift of objects from one frame to another by finding the amount of shift in positions of the centroid. We propose a χ2-test-based local histogram matching scheme for detecting moving objects from complex scenes from low illumination environment and objects that change size from one frame to another. The proposed scheme is successfully applied for detecting moving objects from video sequences captured in both real-life and controlled environments. It is also noticed that the proposed scheme provides better results with less object background misclassification as compared to existing techniques.

In this paper we first introduce the concept of classification confidence in fuzzy rule-based classification. Classification confidence shows the strength of classification for an unseen pattern. Low classification confidence for an unseen pattern means that the classification of that pattern is not very clear compared to that with high classification confidence. Then we focus on the minimum classification confidence for fuzzy rule-based classifiers using the classification confidence. The minimum classification confidence represents the worst classification among given training patterns. Some discussion on assigning a weight to training pattern is given to show that cost-sensitive fuzzy rule-based classifiers are advantageous for producing a large minimum-confidence classifiers. A series of experiments are done in order to show that reasonable classification boundaries can be obtained by cost-sensitive fuzzy rule-based classifiers if appropriate weights are assigned to training patterns.

In this paper, we present a novel algorithm for moving object detection and tracking. The proposed algorithm includes two schemes: one for spatio-temporal spatial segmentation and the other for temporal segmentation. A combination of these schemes is used to identify moving objects and to track them. A compound Markov random field (MRF) model is used as the prior image attribute model, which takes care of the spatial distribution of color, temporal color coherence and edge map in the temporal frames to obtain a spatio-temporal spatial segmentation. In this scheme, segmentation is considered as a pixel labeling problem and is solved using the maximum a posteriori probability (MAP) estimation technique. The MRF-MAP framework is computation intensive due to random initialization. To reduce this burden, we propose a change information based heuristic initialization technique. The scheme requires an initially segmented frame. For initial frame segmentation, compound MRF model is used to model attributes and MAP estimate is obtained by a hybrid algorithm [combination of both simulated annealing (SA) and iterative conditional mode (ICM)] that converges fast. For temporal segmentation, instead of using a gray level difference based change detection mask (CDM), we propose a CDM based on label difference of two frames. The proposed scheme resulted in less effect of silhouette. Further, a combination of both spatial and temporal segmentation process is used to detect the moving objects. Results of the proposed spatial segmentation approach are compared with those of JSEG method, and edgeless and edgebased approaches of segmentation. It is noticed that the proposed approach provides a better spatial segmentation compared to the other three methods.

In this paper, two adaptive thresholding schemes have been proposed. These two schemes are based on adaptive selection of windows based on the proposed window merging and window growing. Windows are selected based on the entropy and feature entropy criterion. PGA and MMSE based segmentation schemes have been proposed to segment the windows selected a priori. The efficacy of the proposed approaches have been compared with the Huang's pyramidal window merging approach. It is found that the proposed approaches exhibited improved performance in the context of accuracy of segmentation.

In this paper, we propose a context-sensitive technique for unsupervised change detection in multitemporal remote sensing images. This technique is based on a modified Hopfield neural network architecture designed to model spatial correlation between neighboring pixels of the difference image produced by comparing images acquired on the same area at different times. Each spatial position in the considered scene is represented by a neuron in the Hopfield network that is connected only to its neighboring units. These connections model the spatial correlation between neighboring pixels and are associated with a context-sensitive energy function that represents the overall status of the network. Change detection maps are obtained by iteratively updating the output status of the neurons until a minimum of the energy function is reached and the network assumes a stable state. A simple heuristic thresholding procedure is presented and adopted for initializing the network. The proposed change detection technique is unsupervised and distribution free. Experimental results carried out on two multispectral and multitemporal remote sensing images confirm the effectiveness of the proposed technique

The objective of this paper is to utilize the extracted features obtained by the wavelet transform (WT) rather than the original multispectral features of remote-sensing images for land-cover classification. WT provides the spatial and spectral characteristics of a pixel along with its neighbors, and hence, this can be utilized for an improved classification. Four classifiers, namely, the fuzzy product aggregation reasoning rule (FPARR), fuzzy explicit, multilayered perceptron, and neuro-fuzzy (NF), are used for this purpose. The performance is tested on multispectral real and synthetic images. The performance of original and wavelet-feature (WF)-based methods is compared. The WF-based methods have consistently yielded better results. Biorthogonal3.3 (Bior3.3) wavelet is found to be superior to other wavelets. FPARR along with the Bior3.3 wavelet outperformed all other methods. Results are evaluated using quantitative indexes like beta and Xie-Beni

This paper describes and analyzes the aggregation pheromone system (APS) algorithm, which extends ant colony optimization (AGO) to continuous domains. APS uses the collective behavior of individuals that communicate using aggregation of pheromones. Two variants of APS are considered: the existing generational APS and the proposed steady-state APS. Both variants of APS are tested on several common unimodal and multimodal problems and their performance on these problems is analyzed with different parameter settings. The results indicate that using a steady-state evolutionary model improves the performance of APS on both unimodal as well as multimodal problems and that the performance of APS is relatively robust with respect to its parameter settings.

A fuzzy rule based cloud classification scheme is proposed to estimate the cloud cover using Meteosat 5 visible imagery over the Indian subcontinent and the Indian Ocean. The technique, which is a supervised one, automatically finds out some human understandable rules (fuzzy rules) for classifying each pixel into one of the three possible classes: clear sky, cloudy sky and partially cloudy sky. In this regard, temporal and spatial properties of the data are being explored. The scheme is tested on images other than the training image(s) and the performance is found to be quite satisfactory.

In this article, we propose a new type of genetic algorithm (GA), the forking GA (fGA), which divides the whole search space into subspaces, depending on the convergence status of the population and the solutions obtained so far. The fGA is intended to deal with multimodal problems that are difficult to solve using conventional GAs. We use a multipopulation scheme that includes one parent population that explores one subspace and one or more child populations exploiting the other subspace. We consider two types of fGAs, depending on the method used to divide the search space. One is the genotypic fGA (g-fGA), which defines the search subspace for each subpopulation, depending on the salient schema within the genotypic search space. The other is the phenotypic fGA (p-fGA), which defines a search subspace by a neighborhood hypercube around the current best individual in the phenotypic feature space. Empirical results on complex function optimization problems show that both the g-fGA and p-fGA perform well compared to conventional GAs. Two additional utilities of the p-fGA are also studied briefly.

An application of Genetic Algorithms (GAs) to evolve Hopfield type optimum neural network architectures for object extraction problem is demonstrated. Different optimizing functions involving minimization of energy value of the network, maximization of percentage of correct classification of pixels (pcc), minimization of number of connections of the network (noc), and a combination of pcc and noc have been considered. The noc value of the evolved (sub)optimal architectures is seen to be reduced to two-third of that required for the fully connected version. The performance of GA is seen to be better than that of Simulated Annealing for this problem.

A large fraction of studies on genetic algorithms (GAs) emphasize finding a globally optimal solution. Some other investigations have also been made for detecting multiple solutions. If a global optimal solution is very sensitive to noise or perturbations in the environment then there may be cases where it is not good to use this solution. In this paper, we propose a new scheme which extends the application of GAs to domains that require the discovery of robust solutions. Perturbations are given to the phenotypic features while evaluating the functional value of individuals, thereby reducing the chance of selecting sharp peaks (i.e., brittle solutions). A mathematical model for this scheme is also developed. Guidelines to determine the amount of perturbation to be added is given. We also suggest a scheme for detecting multiple robust solutions. The effectiveness of the scheme is demonstrated by solving different one- and two-dimensional functions having broad and sharp peaks.

The robustness of neural network (NN) based information processing systems with respect to component failure (damaging of nodes/links) is studied. The damaging/component failure process has been modeled as a Poisson process. To choose the instants or moments of damaging, statistical sampling technique is used. The nodes/links to be damaged are determined randomly. As an illustration, the model is implemented and tested on different object extraction algorithms employing Hopfield's associative memory model, Gibbs random fields, and a self-organizing multilayer neural network. The performance of these algorithms is evaluated in terms of percentage of pixels correctly classified under different noisy environments and different degrees and sequences of damaging. The deterioration in the output is seen to be very small even when a large number of nodes/links are damaged.

Abstract-The feedforward multilayer perceptron (MLP) with back-propagation of error is described. Since use of this network requires a set of labeled input-output, as such it cannot be used for segmentation of images when only one image is available. (However, if images to be processed are of similar nature, one can use a set of known images for learning and then use the network for processing of other images.) A self-organizing multilayer neural network architecture suitable for image processing is proposed. The proposed architecture is also a feedforward one with back-propagation of errors; but like MLP it does not require any supervised learning. Each neuron is connected to the corresponding neuron in the previous layer and the set of neighbors of that neuron. The output status of neurons in the output layer is described as a fuzzy set. A fuzziness measure of this fuzzy set is used as a measure of error in the system (instability of the network). Learning rates for various measures of fuzziness have been theoretically and experimentally studied. An application of the proposed network in object extraction from noisy scenes is also demonstrated. Index Terms- Image processing, object extraction, fuzzy sets, neural networks, multilayer perceptron. I.

Object extraction algorithms with a Neural Network (NN) are described. The objective function to be minimized for object extraction from a scene is shown to be similar to the expression of energy of a neural network. A modified version of Hopfield's Neural Network model is used here. The weights and input biases are given in such a way that the network self-organizes to form compact clusters. Both the discrete and the continuous dynamics of the network have been used for this purpose. Performance of the proposed methods has been compared with that of the relaxation technique.