The network of roads is the only component of a nation that connects its citizens. It also has a significant influence on its economy. Bangladesh invests a huge amount of money in measuring and maintaining the quality of this road network. However, it is time-consuming and needs a lot of manpower. The whole process can be replaced if the measuring of road quality is done by an automation system. In this paper, we describe a proposal to develop a web-based automation system for road quality measurement (RQM) of Bangladesh highways using satellite images. The complete automation system takes satellite road images from the user and shows the result of which class it belongs to by classifying it through the segmentation and classification AI models [1], [2]. The system will use only Satellite images of the roads. No physical survey or manual process is needed. This system will reduce the time, cost, and requirement of a lot of manual work significantly.

Roads are the connection network of a country that upholds the economy. Road quality needs to be measured and repaired from time to time to keep this connection stable. The Roads and Highways Division of Bangladesh spends a huge of money with exorbitant time and labor for this purpose. An automation system using satellite images for road quality measurement of Bangladesh highways can be the best solution that would be much more efficient than the existing manual process and may reduce the time and cost significantly. An efficient satellite image Database is crucial for this intelligent system. No significant research has been done in this sector and no suitable dataset is available in this domain yet. Previous work has been done for a database creation that has many limitations and worked for only straight roads. Curve and covered road detection methods were absent [1]. In this paper, a deep learning-based system that extracts and segments the road from satellite images is proposed for the national highways of Bangladesh. It can be used to classify roads afterward into the automation system. Satellite images come with other objects together rather than only roads. It is needed to segment the roads from the satellite images based on the International Roughness Index (IRI) [2] [3]. Google Earth Pro desktop application is used to collect the satellite images at their highest resolution of 8192x4639 and created our dataset according to the Road Roughness Survey Results of Bangladesh [3]. The total dataset is divided into the training and testing set with a ratio of 1:6. To build the proposed system, nine different deep learning models Deeplabv3, DeeplabV3plus, MAnet, LinkNet, Unet, FPN, Unetplusplus, PAN, and PSPNet are used. First, the training dataset is used for manual segmentation, and the augmentation process is performed on the segmented data before training the model. Then the nine deep learning models were experimented with in different train:validation ratios as 60:40, 70:30, and 80:20. The experimental results show that DeeplabV3plus gained better among all the models maintaining more than 97% accuracy in all ratios. In addition, it provides better extraction than other models in visual inspection.

In this work, we proposed a monitoring system for the road quality measurement of Bangladesh using satellite imagery. The proposed system will enable a much larger scale and much lower costs than are achievable with existing methods. For this task, no dataset was available in this domain. We have collected data and created our dataset. We used two sets of data: a set of road quality management IRI (International Roughness Index) values as our ground truth data, and the corresponding set of satellite images to create our dataset. These data are used to build a model that can automatically measure road quality with the help of different kinds of feature extraction techniques and deep learning architectures. We tried to train the proposed model using two deep learning architectures: Vgg16 and ResNet50. We got the best result in ResNet50 with modification in a 70:30 split ratio where validation accuracy is 70%.

Satellite imagery is a valuable data repository for detecting, quality measurement, and feature mapping. Among all the types of expenditure, preventive maintenance of the road network is the most economical and beneficial. Analyzing the existing architectures for road quality measurement using satellite imagery and comparing the multiple models, we have concluded that Deep Learning models work well rather than the Machine Learning model. To validate our hypothesis, we have carried out a comprehensive study on the existing papers covering available datasets, preprocessing techniques, segmentation, feature extraction, and classification methods. Among the learning models, SqueezeNet which is a compressed version of AlexNet has achieved the optimal performance in determining road class labels in the publicly annotated datasets. Moreover, in terms of road quality measurement, an ensemble approach combining Auto-encoder with LSTMs outperformed in holdout dataset.

In this work, we used MODIS (Moderate Resolution Imaging Spectroradiometer) data MOD13Q1 and MYD13Q1 images to improvise the existing rice model[1] where MODIS NDVI threshold value was used for calculating Boro rice areas in the Sylhet Haor region of Bangladesh. A new algorithm has been proposed here for finding the improvised NDVI threshold value with the Brute Force technique. In total 126 NDVI images have been accumulated and processed in this work. First, a Boro rice model implemented for Sylhet Haor region and ground truth data has been used from Bangladesh Bureau of Statistics (BBS) database for evaluation. The improvised model's accuracy is compared with the previous model. The proposed model shows better accuracy with 2.5% improvement in the Boro rice area prediction for Sylhet Haor region. A new rice model using these improvised thresholds are extended for the whole regions of Bangladesh. The results show better accuracy for all regions. The dataset is used to develop a satellite-based Boro rice area monitoring system. Moreover, it may help in forecasting the Boro rice yield.

In this work, we developed a satellite-based Boro rice monitoring system which will be capable of collecting data on rice production area of whole Bangladesh along with a graphical statistics of rice production area every season. In this monitoring system, Bangladesh has been divided into 7 regions and the production area of Boro rice for these 7 regions has been calculated using only the MODIS (Moderate Resolution Imaging Spectroradiometer) satellite data. Our system does not require any ground truth data. The principal approach requires remotely detected information from MODIS at 250m250m spatial resolutions gained more than two distinct time periods: sowing season which is from December 1 to January 10 and growing season which is from January 11 to April 10 throughout the year (Boro Seasons). Total of eight years 2008-2015 MODIS data have been accumulated and an improvised Normalized Difference Vegetation Index (NDVI) threshold of the Boro rice model has been used for rice area extraction. This monitoring system has some features like region-wise and area-wise information, search options, improvement/deterioration analysis, and rice area extraction map view. The rice monitoring system can be used by the Department of Agriculture of Bangladesh or Food ministry for the collection of accurate and up-to-date information on the country's rice production. It will also be able to provide information on Boro rice production readily accessible to decision-makers and government agencies to help respond better during and after natural disasters or in any need. Moreover, we believe that our developments of forecasting the Boro rice yield would be useful for the decision-makers in addressing food security in Bangladesh.

Remote sensing satellite images have not been thoroughly studied in greenness analysis of Bangladesh. The purpose of this study is to create a systemic database for greenness analysis of Dhaka Division of Bangladesh. Here, we developed a web-based greenness analysis tool using Remote sensing of MODIS (Moderate-resolution Imaging Spectroradiometer) satellite images. MODIS NDVI (Normalized Difference Vegetation Index) is the most popularly used vegetation index in Remote sensing to calculate the amount of greenness. Fifteen years MODIS NDVI images from 2001-2015 are used to extract greenness and accumulated the data of 13 Districts of Dhaka Division to create the database. Using this database and performing algorithms and methodology, this study has found and extracted each green pixels from the original images which is used for analysis. A web-based greenness analysis tool is developed here to display this study of greenness analysis of entire Dhaka. This tool allows mainly four options like Time Series Analysis, Monthly Single District Greenness Comparison, Yearly District Based Comparison and Comparison of two District of Dhaka Division at Chosen Date. These options show how greenness changes yearly among the districts and if they are related to each other and represent the results in graph, bar chart plot and map view. The study has shown us that Dhaka division, and by a larger extent Bangladesh, is an agricultural country and most of the green is based on agricultural crops as it shows increase of green during the period of cultivation for crops.

The aim of this study is to construct a rice yield estimation model for Bangladesh. In this study, Moderate Resolution Imaging Spectroradiometer (MODIS) Normalized Difference Vegetation Index (NDVI) images have been used. The MODIS NDVI images and ground truth data are acquired for the years 2011 to 2016. Since Bangladesh is divided into 8 divisions, several regression models are applied to predict rice yield for each division rather than a single model for the entire country, in order to get improved result in rice yield prediction. Firstly the rice field area is predicted by using NDVI threshold values. An improvised algorithm has been implemented to determine the NDVI threshold values. Four regression models (Linear, Ridge, Lasso, Decision Tree) are performed to estimate total Boro production of each district of Bangladesh. Among the regression models, maximum R2(co-effiecient of determination) values of 0.492, 0.790, 0.899, 0.891, 0.848, 0.942, 0.777 and 0.848 are acquired for Barisal, Chittagong, Dhaka, Khulna, Mymensingh, Rajshahi, Rangpur and Sylhet divisions respectively. Ridge regression worked better for Barisal and Chittagong divisions. For Mymensingh and Rangpur divisions Lasso regression performed the best. Decision Tree regression worked best for the four other divisions.

Remote sensing satellite images have not been thoroughly studied in agricultural sector of Bangladesh.This study demonstrates a rice model using the vegetation index (VI) NDVI (Normalized Difference Vegetation Index) for predicting Boro rice areas in the Haor region of Bangladesh. Existing rice models are not compatible for the latest version of MODIS data. A new rice model with a better range of NDVI (Normalized Difference Vegetation Index) threshold has been proposed for MODIS version 6 and 5 data which provides better results of Boro rice area calculation also when evaluated with ground truth area the error rate reduced from 8.85% to 7.58% on version 5 and from 27.70% to 4.98% on version 6 estimation respectively.

This paper demonstrates an approach to develop a prediction based model for forecasting Boro rice areas in the haor region of Bangladesh. Forecasting the rice areas can contribute in creating a centralized monitoring system for planning effi-cient storage and proper utilization methods. This leads to the development of proposing a new vegetation index (VI). The approach considers a new vegetation index combining NDVI (Normalized Difference Vegetation Index), EVI2 (Enhanced Vegetation Index 2) and OSAVI (Optimized Soil-Adjusted Vegetation Index) for latest version MODIS (version-6) data. The method will forecast total Boro rice areas at the beginning of the Boro season (Dec-Jan) which is more than 3 months earlier from harvesting time without using any ground truth data. 3 Dimensional plotting method and k-Nearest Neighbor classifier have been used on only sowing period (Dec-Jan) data to predict Boro rice pixels. Our new VI has achieved an accuracy of 72%, recall 0.7020, precision 0.4183 and F1score 0.5175.

Many image fusion techniques have been developed. However, most existing fusion processes produce color distortion in 1-m fused IKONOS images due to nonsymmetrical spectral responses of IKONOS imagery. Here, we proposed a fusion process to minimize this spectral distortion in IKONOS 1-m color images. The 1-m fused image is produced from a 4-m multispectral (MS) and 1-m panchromatic (PAN) image, maintaining the relations of spectral responses between PAN and each band of the MS images. To obtain this relation, four spectral weighting parameters are added with the pixel value of each band of the original MS image. Then, each pixel value is updated using a steepest descent method to reflect the maximum spectral response on the fused image. Comparison among the proposed technique and existing processes [intensity hue saturation (IHS) image fusion, Brovey transform, principal component analysis, fast IHS image fusion] has been done. Our proposed technique has succeeded to generate 1-m fused images where spectral distortion has been reduced significantly, although some block distortions appeared at the edge of the fused images. To remove this block distortion, we also proposed a sharpening process using a wavelet transform, which removed block distortion without significant change in the color of the entire image.

Abstract- Forest Fire has serious economic implications: destruction of habitats, forest damage, costs of fire fighting and so on. Nowadays it is very important and sensitive issue in Russia and Southeast Asian region since a large scale fire occurs frequently. A huge amount of exhaustion of carbon dioxide by the forest fires thought to be a cause of global warming. A wide range monitoring by remote sensing satellite is indispensable for the grasp of the fire occurrence situation. The Advanced Very High Resolution Radiometer (AVHRR) flown on the NOAA satellite series is one of the best systems for fire monitoring due to the combination of a very good temporal resolution of several images a day. The forest fire analysis using this weather satellite NOAA has been done by various researchers. However a lot of problems have been left and existing fire detection methods are insufficient. Moreover, a real time fire detection method is necessary for early warning and early detection of fire for fire fighting. In this work a new fire detection method using NOAA AVHRR images have been constructed.

It is important to judge the presence of the snow for prevention of the snow damage to which crops and livestock suffer. However, because of the few meteorological observation point, enough information is not obtained in the plateau region like Mongolia. In this research, the snow area is classified only by the NOAA/AVHRR image assuming application in Mongolia, using three dimensional histogram, and using neither the specialist's judgment nor the grand truth data. The validity of this method is examined compared with snow information that the specialist classified, as the preliminary research in the Tohoku region in Japan. In this method, snow data inside the NOAA images are collected by visual observation, and the snow database is constructed. However, because the image with the snow was few, the data base of an enough size was not able to be constructed. Then, the form of a snow category is extracted using the advantage in which three dimensional histogram was able to be analyzed interactively, and a snow area was classified. As for verification, misclassification rate is that the difference between classified snow area by this method and by specialist's method is divided by the specialist's classification. Misclassification rate became about 50% as a result. Furthermore, misclassification area found that it concentrated on the boundary of the snow. In the boundary, snow is very vague with a resolution of NOAA/AVHRR. When 1 pixel is permitted from the boundary, misclassification rate became about 25%. And when three pixels permitted, it became about 11%. From the above, snow could be classified in the considerable precision by using neither specialist's judgment nor ground truth data. It is applicable in the area with few meteorological observation points like Mongolia, by using this method. The possibility of prevention of damage, such as early precaution of the snow damage, was able to be found.

We have constructed a real time fire detection system by using NOAA/AVHRR for Northeast Asian region. This system has composed of the satellite data received function, image processing function, burned area calculation function and Web opening function. This system tried application from April, 2004

We proposed a high accuracy calculating the scar area by using NOAA AVHRR images. It is based on temporal analysis for fire detection whole a month. The result is plotted on clear sky image map. NOAA AVHRR image is projected by geometric correction to 1.1 km per pixel. Therefore, one plotted fire area is 1.21 km2. It is accumulated every 10 days in a month

Geometric correction methods for NOAA AVHRR image include the GCP (ground control point). Recently, many discussions about the accuracy of geometric correction have been held and many high accuracy geometric correction algorithms for NOAA AVHRR images have been proposed. However, they are not suitable for the time series analysis that processes a large amount of image because they need the manual procedures for the creation of GCP. In this research, we paid attention to this point and developed the completely automatic GCP creation method. The processing time of this method was about 1 second. So, it was confirmed that GCP was generated at very high speed.

For the high resolution satellite image database obtained from IKONOS and Quickbird, we proposed a design including category classification system. These images are essentially different from the image NOAA AVHRR, Landsat TM, Spot, etc., because obtained image is understood like an aeronautical photograph. This system has a knowledge collection and high quality training function to classify the category.