ACM Based ROI Extraction for Pedestrian Detection with Partial Occlusion Handling

Pedestrian detection in video surveillance systems is an integral part of Advanced Driver Assistance Systems (ADAS). In this paper, a new method for efficient pedestrian detection is proposed. The proposed method uses ACM (Active Contour Model) for efficiently locating pedestrian position in each video frame and thereby speeding up the detection time. This method uses a combination of HOG (Histogram of Oriented Gradients) and LBP (Local Binary Patterns) as features for training a two level linear SVM (Support Vector Machine). The proposed method handles partial occlusion using a two-level SVM classifier and eliminates multiple detection using Non Maximum Suppression (NMS) algorithm. The performance analysis is done using INRIA Person dataset and CVC Partial Occlusion dataset; and it is found that the proposed method gives promising results in terms of detection accuracy and detection speed.