Detection of Objects and Activities in Videos using Spatial Relations and Ontology Based Approach in Video Database System

The psychovisual analysis and drilling in detail for extracting the content is key for understanding objects & activities in video. The goal to retrieve some desired content from video database in economical and semantic way is challenging one. Typical applications that needs modeling and extracting semantic video content are useful for systems like police investigation, video-on-demand systems, intrusion detection, border watching, sport events, criminal investigation systems etc. In current scenario most of the techniques are inefficient, subjective, costing in time & limiting the querying capabilities. So there is urgent need for effective method to bridge the degree of gap between low-level representative options and high-level linguistics content. This is achievable upto certain degree by specifically focusing on segmentation & classification aspect of video retrieval process. In each of this area an inclination is booted to emphasize the importance of understanding domain-specific characteristics. In segmentation, the thought of an estimable scene as a bit of visual data that exhibits semi-permanent characterization with visual properties connection establishment is done. Here, linguistics content extraction system that permits the user to question and retrieve objects, events, and concepts that are measured & extracted automatically is proposed. The classification & segmentation techniques that exploits domain-specific special & structural constraints in addition to temporal shift models is considered for jet coaching, auto-racing & train video for detection of objects & activities.