Motion Detection for Content Based Video Retrieval

Abstract

One of the challenges faced in video retrieval is to accurately detect the objects in motion. The features of the moving objects from the input query clip are extracted and compared with the features of the objects present in the videos stored in the database. If the similarity of the features is matched up to the threshold level, then it indicates the objects are similar. An abstract difference technique is used to identify the moving objects in the videos in this paper's proposal. Noise filtering comes after this one.