Content based Video Querying Technique for Video Retrieval and Video Making from Large Video Compilation

Content based video querying and video matching systems are popular in the recent technology. The content based video querying takes a sample video clip as an input query and performs the searching operation in the collection of videos which are stored in the video database. This proposal, introduces a novel content-based video matching and copy elimination system that finds the most relevant video segments from video database based on the given query video clip. For effective video copy elimination based on the feature extraction the proposed system applies the scheme names as Dense SIFT_OP (DSIFT_OP). This performs the feature extraction, copy elimination and effective query matching from the video collections. This thesis overcomes the problem of video frame mining based on effective Meta information’s and semantic similarity measures. The semantic similarity contains both textual and visual similarity measures. According to the discovered features and patterns, the query frame can obtain a set of relevant video frames in the refinement process. The proposed approach robustly identifies the duplicate frames and alignsthe extracted frames, which containing the significant spatial and temporal differences. Based on the feature extraction algorithm and semantic feature identification this applies a motion matching alignment scheme image alignment and video making with extracted clips in the large video database framework. For image analysis and synthesis the image information is transferred from the nearest neighbors to a queryimage according to the distance. This framework is demonstrated through concrete applications, such as motion field prediction and pattern analysis from a single image, pattern synthesis via object transfer, image registration and object recognition. The proposed sequence of object and distance finding yields better result for video making and video copy elimination