Aiming at the problem of water environment protection using fish motion tracing, the fish motion monitor research is proposed based on the video algorithm. It can obtain trajectory and velocity of the fish by means of segmentation and tracing of the fish motion video, which lie a theory foundation for the fish to participate in the environment pollution research (through the comparison of some parameters of the fish motion in different polluted environment, to carry out quantitative analysis of the water pollution degree). The algorithm uses marker multi-scale watershed method for fish motion segmentation, then tracks the fish motion video with the improved weighted-Hausdorff. Finally, in order to accommodate the shape change of the fish in the moving process, introduce the Euclid norm into multi-value images as the restriction to accomplish the update of the tracing model. The experiment shows that the proposed method presents the state of stable tracking, the position shift in the tracking of the successive 100 frame is within a pixel, and the tracing velocity difference is within 0.12 pixels, which can segment and track the fish motion object fleetly and precisely.