This work describes decomposition and synthesis techniques applied to remote sensing image analysis and focuses on the need for analyzing a series of segments in accordance with the variance of scenes. The algorithms described here take two unique approaches. The blocking image method is used for constructing a series of segments. The length function of area is defined that determines the strength of the similarity between a pair of lines. Some tests are also described

and the results indicate that the methods work well for a variety of images.