Low and moderate resolution synthetic aperture radar (SAR) images are used to extract texture features with different texture window sizes by the gray-level co-occurrence matrix and construct the difference images. Otsu threshold segmentation method is combined to obtain the change image. The experimental results show that，as for the detection of areas with simple surface features and obvious changes, the low and moderate resolution SAR images can meet the high detection accuracy demand by using the mean-value texture feature in combination with the corresponding texture window.