Hybrid Video Segmentation with Feature Extraction using Anisotropic Diffusion

In recent years, video segmentation is considered as the major research area for digital storage. Video segmentation is different from the image segmentation. Video segmentation is a challenging problem. In video segmentation, for a given image the segmentation achieved should be related to the previous image that belongs to the same shot. In our proposed video segmentation technique, at first the similar shots in the video are segmented by using conversion of frames using shot segmentation. Next for each shot the track frames are collected with the help of the extracted objects in every frame by using anisotropic diffusion. Effective video segmentation results are obtained in the proposed hybrid video segmentation technique by performing intersection on the segmented results provided by both the frame difference method as well as consecutive frame intersection method. Here we utilized the anisotropic diffusion method for the object extraction from the video shots for segmentation. Therefore our proposed technique is evaluated by varying video sequences and also the efficiency is analyzed. KeywordsVideo segmentation, Anisotropic diffusion, frame difference method, intersection method.