A new texture-based change detection approach is presented that involves texture segmentation to describe the influence of different types of the surficial congiguration. Two images of the same position but different time are concerned. The former image, the reference image is divided into a series of textures standing for different surficial configurations, and its is assumed that the gray value of the above textures is modeled as Gaussian distributions. Through comparing the latter image, the test image, with the former, the pixel point that deviates significantly from the distribution of the texture is the singular changing point. The algorithm analyzes the statistic property of the pixel value over segmented textures, not sliding windows, so it can detect changes of any size or shape. Experiments are carried out to prove the validity of the new approach.