摘要
When fish schooling behavior is studied in a laboratory environment, video multi-tracking systems must automatically and correctly track individual fish. However, most multi-tracking systems do not perform well in this regard. To resolve this problem we develop a novel method for tracking fish in schools. The tracking process searches the state of the target fish by using the previous state of that fish, limiting the search to the attention region centered in the target fish. The attention region is then updated according to the new state of the target fish. We apply this method to track fish swimming in small schools, and demonstrate it to achieve up to 99% accuracy. Our method might find application in water quality monitoring.
- 出版日期2016
- 单位浙江工业大学