Underwater Moving Object Detection Based on Codebook Model

Underwater video intelligent monitoring system extracts moving objects in a complex and dynamic scene with the impact of wave, illumination changes and reflection, while needs real-time character. To solve these situations, underwater moving object detection algorithm based on codebook model was presented. The algorithm creates a background codebook for each pixel according to color distortion and brightness distortion, and then extracts the moving objects using the current pixel to match with code words in the codebook. To adapt the complex changes of scene, the algorithm updates adaptively the background codebook model using the sample value of pixel of current image. The experiments show that the algorithm can extract the complete moving objects by overcoming the impact of underwater environment and inhibiting noise. And the algorithm has perfect real-time character.