Water quality monitoring using multi-object tracking algorithm

HU Jiang-long， FANG Jing-long， WANG Da-quan
（Institute of Graphics and Image， Hangzhou Dianzi University， Hangzhou 310018， China）

Abstract： Aiming at the problems of tracking the fishes in live-fish water quality monitoring system，a video-based multi-target tracking algorithm was proposed. The initial background of the experimental fish tank was built using a statistical background modeling method，and the background was updated during target detection and tracking in real-time. On this basis，background subtraction method and adaptive image binarization were used to achieve the extraction of fish moving target，and the connected component analysis was used to extract fish size，mass and other characteristics of value，and finally the Kalman predictor for the motion tracking based on characteristics was applied to achieve trajectory tracking of fish，extracting fish's trajectory. It is achieved that providing a reliable trajectory data to the follow-up early warning of water quality eventually. The data of the tracked trajectory is an important basis for analysis of water quality in the monitoring system. The results show that the algorithm，with the requirements of real-time and accuracy in the system，could achieve trajectory of fish tracking accurately and quickly.
Key words： water quality monitoring；multi-object video tracking；image segmentation；computer vision