ABSTRACT
The selection of statistically homogeneous pixels (SHPs) has a great significance in interferometric applications. The essence of the measurement of statistical homogeneity lies in the assessment of the degree of similarity between two pixels. Various amplitude-based strategies have been proposed for selecting SHPs. However, the detection rates of these methods are usually unsatisfactory in the case of small data sets. To overcome this limitation, in this work, SHPs are selected based on the similarity test of the covariance matrix. In addition, the covariance matrix of each pixel is estimated by an adaptive M-estimator, which has a robust performance in heterogeneous scenes. In the implementation process, image segmentation is introduced to guide the selection of SHPs in each search window. The feasibility and effectiveness of the proposed method are demonstrated using experiments on simulated and real data.