Human Facial Expression Based Video Retrieval with Query Video Using EBCOT and MLP

Abstract

We can find a massive demand for systems that retrieve video with a query as image or video since the text-based retrieval system (text, images, video) was obsolete and the content-based video retrieval oriented. For many applications, ‘it is essential for retrieving videos from the extensive database with image and video query for content linking and brand monitoring. In this paper, a new video retrieval scheme developed for retrieving videos from an extensive database with a video query. The process uses EBCOT (Embedded Block Coding with Optimization Truncation) coding for feature extraction and MLP (Multilayer Perceptron) for classification and retrieval. The process uses videos with human facial expression as query and related videos with similar facial expression can be retrieved. The process carried out with three phases i)facial expression recognition ii)frame indexing iii)video indexing and retrieving. This methodology does video retrieval process in which video gives as a query for which facial expression has been recognized videos in the database are indexed with the corresponding expression and finally retrieval has been performed thus in this method retrieval scalability has been improved significantly.