A Biological Hierarchical Model Based Underwater Moving Object Detection

Algorithm 1

Hierarchical background modeling algorithm.

Input: Underwater image sequence

Step  1. Segment each frame of input underwater image sequence into multiple nonoverlapped

   subblocks;

Step  2. Extract intensity features of each subblock, 𝜈=𝜇_ℎ𝑡,𝜇_ℎ𝑏,𝜇_𝑙𝑡,𝜇_𝑙𝑏;

Step  3. Establish the background model based on each subblock to distinguish background and

   rough object;

Step  4. Update parameters for the background model;

Step  5. Extract the texture features of each pixel in rough object region, LBP_𝑃,𝑅;

Step  6. Build background model based on each pixel to obtain object contours.

Output: Results of moving objects detection