An accurate algorithm for land surface changes detection based on deep learning and improved pixel clustering using SAR images

Synthetic Aperture Radar (SAR) was developed to map the terrain without the use of large antennas. Combined array is one of the radar methods that is applied from an aircraft or a space platform and in which an effective aperture of the antenna is created in a combined manner. The images obtained by these radars are very accurate. On the other hand, the surface of the earth always changes due to various factors. Accurate identification of these changes can be used in many applications, especially in Iraq due to its semidesert structure. In this research, an improved surface change detection algorithm based on morphological transformation, two-stage center-constrained FCM algorithm (TCCFCM) clustering and deep learning is presented. The simulation of the method of this research on MATLAB software shows that the proposed method is fast and inexpensive. The accuracy was 99.7%.