Moving Object Segmentation from Underwater Videos Using Adaptive Collective Background Learning Approach

Countless approaches are studied in the literature for segmentation of moving object in image and underwater video frames. As a basic models, Adaptive background learning and frame Difference BGS plays a very important and is simplest method of segmentation with less computation competency. Adaptive background learning method updates the background by considering the original input image with some threshold value and Frame Difference BGS approach is required to find the absolute difference between the current frame and background frame. Hence, we have proposed an Adaptive Collective Background Learning approach, which is an improvement on the back ground modeling. It is a collective method of background modeling, which updates the background in terms of weightages. The proposed method can effectively resolve the problem of disturbance caused by noise and improves the segmentation results of moving object. Experiment results demonstrate that the proposed method has a significant capacity in suppressing noise in terms of precision and PWC as compared to basic models and it is more efficient than basic models.