An Efficient Method to find Video Objects

Abstract

Modern Computer Technology, together with the proliferation broadcast channels and of video-based survelliance systems, has en-abled us to produce vast amounts of both raw and processed video data. The potential uses of this data are many and varied. Monitoring and mining of the contents of this is already huge, rapidly growing mass of data calls for the development of major computational resources and the development of sophisticated video understanding techniques. Video extraction / video mining is a highly challenging tasks. It deals with the identifying / capturing of object of interest from the video. Some common area where this video mining is useful are: identifying the pitch of ball inside / outside the boundary, captures sachin’s role in the entire match, analyses the winner in the running race, etc. In this article, we pro-pose a suitable technique for capturing the moving object in video using frames. We capture the object and place it in frame in a sequence man-ner. The background in the object is eliminated by some technique. Thus the foreground image is well displayed. After elimination of the background, the objects are compared with each other. Finally, the difference in the frames is captured. Thus the particular object is captured from the video. The objective is to monitor and capture the object in a random manner. The framework for identifying the object is by re-cording through a stationary camera, and captures the sequences of objects. Dynamic objects can be captured by both background elimination and background registration techniques. Thus we propose the suitable methods to mine the object from the video in a more easy way. The background registration method uses background subtraction which improves the adaptive background mixture model and makes the system learn faster and more accurately, as well as adapt effectively to changing environments.

 

Keywords: Background elimination, background registration techniques, modern computer techniques, monitoring and mining of the content, Object identification, stationary background, stationary camera.