Maximally Stable Extremal Regions Improved Tracking Algorithm Based on Depth Image

In order to solve the problem that traditional Camshift algorithm can easily fail to track overlapping targets and multiple similar depth targets, a new improved maximally stable extremal regions (MSER) algorithm is presented in this paper. Firstly, the suspected target contour is extracted and similarity analysis is performed. Secondly, the improved MSER algorithm is used to confirm the target contour and update the similarity library. Finally, combined with the physical properties unique to the depth image and based on the Kalman filter, it is possible to predict the tracking target's moving position. The experimental results show that the real-time performance and recognition rate are improved, and robustness to the situation of target overlap and occlusion is better with the improved MSER algorithm.