Abstract: The multi-beam forward-looking sonar has the advantages of high imaging speed and high resolution. It is an important device for underwater target detection, tracking and monitoring. Aiming at the tracking problem of multi-beam forward-looking sonar, an improved Mean Shift algorithm is proposed. In the algorithm, the classical Mean Shift algorithm is used to locate the inter-frame of the target, and the background subtraction method based on the image sequence is used to segment the moving target. The Mean Shift tracking box is updated according to the position and size of the segmented target, and the tracking model is re-established to achieve accurate localization and tracking of the target through iteration. The experimental results show that the improved algorithm can update the target tracking frame with the size and shape of the target, and the target localization is more accurate. Therefore, the algorithm has the potential to be applied to accurate tracking and localization of underwater targets.