International Journal of Computer Applications (0975 – 8887) Volume 75– No.13, August 2013 1 Object Detection and Tracking using Background Subtraction and Connected Component Labeling Asad Abdul Malik Institute of Information & Technology, Kohat University of Science & Technology, Kohat, Pakistan Amaad Khalil Department of Computer Systems Engineering University of Engineering & Technology, Peshawar, Pakistan Hameed Ullah Khan Department of Computer Systems Engineering University of Engineering & Technology, Peshawar, Pakistan ABSTRAC Digital image processing is one of the most researched fields nowadays. The ever increasing need of surveillance systems has further on made this field the point of emphasis. Surveillance systems are used for security reasons, intelligence gathering and many individual needs. Object tracking and detection is one of the main steps in these systems. Different techniques are used for this task and research is vastly done to make this system automated and to make it reliable. In this research subjective quality assessment of object detection and object tracking is discussed in detail. In the proposed system the background subtraction is done from the clean original image by using distortion of color and brightness. The subtracted image is then tracked using connected component labeling. The proposed system eliminates the shadow and provides 79% accuracy. Keywords Object tracking, detection, Background Subtraction, color distortion 1. INTRODUCTION The detection of a moving object and tracking of different objects in a video or video sequence is a very important task in the surveillance videos, analysis and monitoring of traffic, tracking and detection of humans and different gesture recognitions in human-machine interface [1]. The technique of Object tracking can be explained to be the method of tracking the different number of objects in the video and also the certain directions those objects are traversing in and also to track the entrances to the surveillance site as per the unit time. The sophistication and the complexity of the system determine the resolution of the measurement. This system is often deployed in public places such as shopping malls, metro stations, airports and independent surveillance requests. Different approaches can be used for the surveillance and different technologies used as computer vision, infrared beams and thermal imaging [2]. The reasons for object tracking are many For example People counting in retail stores for intelligence gathering can be regarded as one. This is used for the calculation of the conversion rate and rating of the store by the number of customers to the store rather than the old use of the sales data [2]. The video analysis can be safely stated to be consisting of three steps: 1 st detecting the objects that are interested in, 2 nd the frame to frame tracking of those objects and 3 rd the analysis of the path traversed to analyze their behavior [3]. The agenda of this paper is as follows. Section 2 presents Literature that was studied for the background study and better understanding the topic. Section 3 describes the proposed system in view of the study and the current research. Section 4 shows results of the proposed system of this research. Discussion is done is Section 5 and Section 6 contains the conclusion of the current work. 2. LITERATURE REVIEW 2.1 Object Detection One of the simplest techniques for the object detection is the background subtraction. In this process the observed frame or image is compared with the same scene but with the exclusion of any objects in the scene. The subtraction from the original scene results in the difference of the two images. The difference highlights the areas with significant change and hence identifies the areas of interest [14,4]. The technique for the background modeling can broadly be classified in two categories i) Non-Recursive and ii) Recursive. 2.1.1 Non-Recursive Technique This technique involves the use of storing the previous L video frames. The estimation of the background is done with the temporal variation of each pixel within the buffer [5]. A few non-recursive techniques used commonly are described below. 2.1.1.1 Frame differencing This technique arguably is one of the simplest techniques for background modeling. The video frame used as a background for the frame at time t is the frame at time t – 1 is used [6]. 2.1.1.2 Median filtering This technique defines the background estimation as a median of all the frames in the buffer at each pixel location processed. It is also assumed that for half of the frames in the buffer the pixel stays in background [6,15].