A STUDY ON APPROACHES FOR IDENTIFYING HUMANS IN LOW RESOLUTION VIDEOS

Human detection in surveillance systems is becoming one of the major issues in present scenarios for many applications like human gait appearance, gender classification, security systems etc. Many methods have been proposed for obtaining human information from videos with low resolution. Also it is very much required to protect privacy of individuals in the video. Human detection is more difficult in low resolution videos compared to high resolution videos. In low resolution videos the pixels size is more and also have very less information for identification. Major things to identify are object detection and object classification in low resolution videos. In this paper we performed a detailed study on various detection and classification techniques besides, Background subtraction methods are studied. Basically human beings are classified based on shape, texture and motion based features which are necessarily to be identified. The comparison of existing techniques is presented in this paper. Many datasets are available for performing experimental evaluation and to find new techniques for improving the detection of human in low resolution videos. Some of the datasets are discussed in this paper.