Abstract: River change detection based on remote sensing imagery and vector data is introduced.According to different shapes,rivers are divided into main rivers,lakes and confluent net."Pattern classification-region growing-morphologic processing-edge detection-buffer detection" are employed to main river change detection;"region growing-morphologic processing-polygon area subtracting" are used to lake and "pyramid decomposing and multiply scale template matching-Ziplock Snakes-buffer detection" are applied to confluent net.Experimental results show that schemes mentioned above are feasible.