Gabor filter based change detection in SAR images by KI thresholding

In this paper, automatic change detection (CD) method using Gabor filter is proposed for Synthetic Aperture Radar (SAR) images. Fixed sized Gabor filter is designed using its parameters. CD involves two SAR images; hence a difference image is generated by logarithmic ratio of two images. Dot product is performed between Gabor filter in row vector format and local window of the difference image in column vector format to localize boundaries of the changed area accurately. Kittler- Illingworth thresholding algorithm is exploited to produce a binary change map to discriminate changed and unchanged pixels. Experiments are carried out to validate the effectiveness of the algorithm with and without preprocessing speckle removal filters. Experimental results obtained on SAR images acquired by the ESA Envisat, ERS-2 and ERS-1 satellites show the achievement in terms of quality metrics such as low false alarm rate, high CD accuracy and less computational time. It shows the efficacy of the proposed algorithm when speedy detection is required during natural calamities.