A Novel Technique for Moving Object Detection & Tracking with Optical Flow

Abstract

A novel technique for tracking for tracking of objects in moving state is proposed to done by optical flow analysis which provide accurate results. In this Optical flow analysis, the velocity vectors of the pixels are obtained. As a result, velocity field is analyzed for the frame sequences which are considered and described for short duration of the video. This method doesn’t need to refer the previous frames which reduces the processing time and can be applicable for real time events. This novel algorithm produces better results with repeatability, uniqueness and reliability, compared to previous techniques. This method can be able to perform faster computation and comparison. This can be achieved by performing image convolutions for integral images followed by technical operations in the descriptors and detectors. Finally these methods can be simplified to the needed requirements. This leads to a package of better quality detection followed by description and recognition steps. The paper encircles a brief explanation of the algorithm and the results of simulation are also described in detail.