Change detection in synthetic aperture radar images using spatial fuzzy clustering based on the similarity matrix

Change detection is one of the most important applications of synthetic aperture radar (SAR) imaging in remote sensing, for environmental monitoring, disaster assessment, and urban development. SAR possesses unique advantages: all-weather and day-night operation, but subtle surface changes are difficult to detect. In addition, complex backscatter patterns and speckle noise further complicate the traditional detection methods. This work presents an enhanced method for SAR-based change detection using the spatial fuzzy clustering method based on the similarity matrix. The overall process involves the following preprocessing steps, such as noise reduction, data normalization and image registration for achieving quality data. The similarity matrix is used to calculate the differences between multi-temporal SAR images, and then, spatial fuzzy clustering membership (SFCM) is applied to segment the images into changed and unchanged areas. The iterative optimization of the clustering process gives a notable reduction in loss values from 0.2431 at 100 iterations to 0.0044 at 1,100 iterations. Test results validate the efficiency of the method, which also achieved a testing loss of 11.63 and an accuracy of 98.84%. The results showcase the potential of the proposed method to be used in surface change detection that is robust for overcoming SAR-specific challenges. Advanced techniques of clustering as well as preprocessing demonstrate the importance of advancing the SAR-based change detection process.