Change Detection Classification of Buildings from Satellite Images of Urban Area using Deep Learning Techniques

As the need for analysis of data has emerged it has become important for examining the change detection in detecting the buildings. A very High-Resolution image has a wide range of applications with regards to specifying surface locations within depth. Buildings extraction from VHR has become more popular in many real time applications. It is difficult for a human specialist to manually extract the relevant data, thus building detection and feature extraction from VHR images are critical. In the proposed system, the comparison of built-up areas from datasets of different dates is done. Initially consider a large dataset and then the data is pre-processed for image resizing, image cropping and removal of noise. For this, at first classify the constraints as Class-A: Vanishing, Class-B: Emerging and Class-C: Unchanged. As the count of buildings has been rapidly increasing every day, one can achieve the comparative analysis of built-up areas and determine the type of building, the length, breadth and hence the area from both the datasets and determine the change of class from one to another. The two images before.png and after.png of the same place at different times each are of 1024*1024 pixels. The first half focuses on image processing for urban area classification, while the second portion describes and compares built-up areas.