Urban change detection analysis utilizing multiresolution texture features from polarimetric SAR images

Abstract

This paper presents a multiresolution textural approach to change detection in multitemporal polarimetric synthetic aperture radar images. Several change detection methods utilizing wavelet based features from remotely sensed images have been developed previously. The aim of this paper is to investigate and propose a texture based change detection method that applies curvelet and contourlet transforms on polarimetric synthetic aperture radar (SAR) images. The proposed approach exploits curvelet and contourlet based multiscale decomposition of original images where textural information is extracted at various scales and in different directions in terms of statistical moments and energy to generate feature maps. The L1-norm is used in the proposed method to generate the difference image, which is thresholded using the maximum entropy principle to obtain the final change detection map. The results are compared with the changes detected by the wavelet based textural features, plain texture difference, image difference and log ratio methods. Finally, accuracy assessment and comparative analysis are performed for change maps carried out to review missed changes, false-alarms and the overall accuracy of this method. The results are also analysed statistically with area under ROC curve (AUC). The study found that the proposed method produces results with high change detection accuracy with better edge continuity and greater AUC.