A Method of Object Recognition which Applicable to the Real-Time Processing Using Adaptive Factors

In a video with various environments, background modeling is important for the extraction and recognition of the moving object. The Kumar method represents queue-based background modeling, because it has a fixed period of updating the inspection of a frame, there is a limit to various systems. This paper proposes a decision-making method of the factors responding to adaptation, in accordance with the background models for the period of updating the inspection of the frame and of the factors which decide the queue size for an update inspection. The grouping of which reduces in a processing time. The proposed method can improve the existing techniques of background modeling, which is not appropriate for real-time processing. In addition, the proposed method can recognize to objects more efficiently.