A background reconstruction algorithm based on pixel classification and block analysis for aquatic surveillance system

This paper investigates the challenges faced when reconstructing the background in an outdoor aquatic environment and demonstrates a new background reconstruction algorithm for aquatic surveillance system. First pixels are classified by the hue and intensity values in the HSI color space; then background pixels are determined by a block-based analysis algorithm. The background is updated in different ways due to the background change speed. Experiments indicate the developed algorithm can get robust performance for different hostile environments faced by an outdoor aquatic environment.