ISSN: 2005-4297 IJCA Copyright ⓒ 2020 SERSC International Journal of Control and Automation Vol. 13, No. 2, (2020), pp. 868 - 876 868 Proximity Approach for Object Detection in Video Nilesh Uke, Member IEEE Professor, Department of Computer Engineering Trinity Academy of Engineering, Pune, India , Shailaja Uke Assistant Professor, Department of Computer Engineering SKN SITS, Lonavala, India , Abstract Last decade we are experiencing more applications in video surveillance to address issues related to social needs. As public concern about crime and terrorist activity increases, the importance of public security is growing, and video surveillance systems are increasingly widespread tools for monitoring, management, and law enforcement in public areas. Object detection is a primary concern about all of these applications domains. In this paper, we exploit computer vision methods to detect moving object from video to track in real time as objects encountered in the indoor and outdoor environment. Proximity is a fact of being near to other and justifies closeness. These concepts of object being close to each other is checked while the process of object tracking. System tracks assorted objects against an environment consisting of objects of varying sizes, shapes and colors. Initially background modeling is performed using the function which accumulated the background frames from mean and standard deviation of first N frames. Each significant change in the object appearance thereafter, due to new object, old object disappearance is tracked based on the proximity of the target object. The visual resemblance is determined with respect to the detected object in the previous video frame and the last frame captures. Keywords: Moving Object, Object Tracking, Proximity, Gestalt Law, 1. Introduction Methods for extracting moving objects from videos are being studied extensively by many researchers due to its wide verity of applications. Once the moving object detected; it is being used in many application which includes measuring vehicle traffic [1], motion tracking [2],[3], traffic sign recognition [4]–[6], pedestrian detection [7], [8] , face and logo detection [9]–[11], and drivers drowsiness detection [12]. But in recent years, due to increased demand of intelligent systems and more challenging real world scenes made systems to be more robust to noise in data, abrupt motion or illumination variation, non-rigid or articulated movement of objects, background variation etc. The main difficulty to solve tracking problem is to find correspondence of the same moving objects in different frames of the video. This problem may solved by looking at several aspects of the scene, such as the density and proximity of objects, variable shapes, presence of occlusions etc. The problem is further complicated by several factors such as camera quivering, flawed calibration of the on-board cameras, complex environments, and so on [13]. 2. Background Study Most recently, many research related to visual tracking is being carried out. Stereo vision-based model for multi-object detection and tracking is proposed for surveillance systems [14]. Computer Vision methods and deep convolutional neural networks (CNNs) are seemingly combined in DEP- SEE framework [15] to exploits to detect, track and recognize in real time moving objects observed during moving in the outdoor environment. Earlier we proposed hybrid method of object detection using motion estimation and tracking by parallel Kalman filter [16]. A system [2] is proposes with a unique object detection and tracking system where video segmentation, feature extraction, object detection and tracking are combined perfectly using various features. 2.1 Visual Perception and Proximity