Histogram of Oriented Gradients with Kalman Filter approach for object Detection and Tracking
Object detection and tracking has a challenging task in computer vision applications such as traffic monitoring, video surveillance, robot vision military guidance. This paper proposes detection and tracking of object using Histogram of oriented gradient (HOG) integrated Kalman Filter. HOG descriptor is used to describe a target vehicle and using Kalman filter detected object has been tracked. After performing foreground and blob detection on the acquired datasets, feature extraction (Histogram of Oriented Gradients (HOG), was carried-out for extracting the feature values from the blob detected image frames. Particle swarm Optimization (PSO)was used for selecting the optimal feature subsets. The experimental outcome shows that the proposed system performs effectively by means of precision, recall, f-measure and accuracy. The proposed system enhances the classification accuracy compared to the existing systems. The computation cost is mainly paid in processes of both feature extraction.
Keywords - Histogram of Oriented Gradients; Particle Swarm Optimization; Kalman Filter