Abstract |
: |
Adept identification of similar videos is an important and consequential issue in content-based video retrieval. Video search is done on basis of keywords and mapped to the tags associated to each video, which does not produce expected results. So we propose a video based summarization to improve the video browsing process with more relevant search results from large database. In our method, a stable visual dictionary is built by clustering videos on basis of the rate of disturbance caused in the pixellize, by the object. An efficient two-layered index strategy with background texture classification followed by disturbance rate disposition is made as core mapping methodology. |