Coherent change detection with complex logarithm transformation on SAR imagery

A satellite-borne SAR (synthetic aperture radar) is quite promising technique for high-resolution geosurface measurement. Recently, the feature extraction method based on the CCD (coherent change detection) has been developed, which detects a small surface change on the geosurface by using the phase relationship between the plural complex SAR images of the same region in different observations. Aiming at fast and accurate detection of the surface change, the logarithm transformation method, has been proposed. This method can determine the appropriate threshold for the change detection, while enhancing the detection probability and suppressing the false alarm rate. However, it does not employ phase information of the estimated coherence function, and its detection probability deteriorates, especially in the case when target has small surface changes. To overcome this problem, a novel transformation index is proposed considering the phase difference of the coherence function. The results from the experiment modeling of geosurface measurement verify the effectiveness of the proposed method, even in the lower SNR (signal to noise ratio) situations.