Abstract—Detecting objects in video sequences is a challenging mission for identifying and; tracking moving objects. Background removal is considered a basic step in detecting moving objects tasks. Dual static cameras placed in the front and rear moving platform gather information which is used to detect objects. Background changes regarding peed and direction of the moving platform, so moving objects so distinguishing moving objects becomes complicated. In this paper, we propose a framework which allows detection of moving objects with a variety of speed and direction dynamically. Object detection technique is built on two levels firstly level apply background removal and edge detection to generate moving areas. Secondly apply Moving Areas Filter (MAF) then calculate Correlation Score (CS) for the adjusted moving area, merging moving areas with closer CS then marking as a moving object. Experiment results are prepared on real scene acquired by dual static cameras without overlap in sense. The results show accuracy in detecting objects compared with optical flow and Mixture Module Gaussian (MMG). An accurate ratio is produced to measure accurate detection moving object. Keywords—Background Removal, Correlation, Mixture Module Gaussian, Moving Platform, Object Detection. I. INTRODUCTION N most video system applications the first step is detecting objects which separate the foreground from the background. Dealing with real time applications requires massive amounts of computing resources, whereas current real time techniques make limiting assumptions about object or platform movement or scene structure. Background removal is an effective method to detect object in stationary backgrounds [1]. Background subtraction provides many false detection points for a dynamic scene. Moving objects detected in a dynamic environment is still a problem under research. Moving object detection is efficiently handled by Adaptive background removal algorithms when a scene is a dynamic background [2], [4]. There are several challenges that must be confronted when detecting moving objects. It is difficult to distinguish between objects and background. First, both of them have to move closer in speed and direction. Second a quick response to dynamical environment is needed. Previous challenges present serious limitations to real time requirement. Object detection is the spatial accuracy and temporal coherency of localized independently moving objects detection [3]. Despite its importance, detection of moving objects in dynamic backgrounds is far from becoming highly Sallama Athab is with the Northampton University, (e-mail: sallama.resen12@my.northampton.ac.uk). Hala Bahjet is with the Technology University, Head Information System branch (e-mail: hala_bahjat@ yahoo.com). accurate [5]. Motion in video acquired from static cameras placed on moving platforms occurs by moving objects of interest and the background. Optical flow is essential in detection of the moving object task. The drawback of the Optical Flow algorithm is time consumption which make it useless in real time video system applications such as traffic, monitoring. Optical flow regularization acts locally, so does not make use of the large region might which have constant motion [6]. Static camera research is now including background removal, object detection and feature tracking techniques. This clearly is not sufficient when a static camera is placed on a moving platform. Changes in frames are caused by combined platform and object motion. This paper distinguishes between two types of movement in sequence frames: the first one is called virtual movement, caused by background changes when on a moving platform such as a road sign or a tree on the side of the road. Second one is called real movement, caused by moving objects with a relative speed to the moving platform. This paper uses the current MAF to isolate virtual movement depending on pixel location and intensity. Real movement is measured by SC. SC measures speed and direction, where change depends on spatiotemporal relation. In this paper object detection algorithms have been built to handle a variety of speeds between platform and objects movement. In this paper the problems to overcome are presented in Section III, proposed technique is presented in Sections IV and V, results and discussion are in Section VI and it is concluded with future work. II. REVIEW Detection of moving objects in a dynamic sense can be considered the lowers level of operation in computer vision to get higher level event analyses, so moving objects detection is an essential task for many computer vision applications such as video traffic management and surveillance monitoring systems [2]. Static cameras in moving platforms capturing occurs at time t this point is moving in time t+1 therefore processing is a function of time. The object of interest is determined by the location filter [7]. Spatial and temporal features based on Pixels are used such as edges and motion information to detect foreground with a scene background that is dynamically changes. Good spatial accuracy is achieved by edges; it is suitable because of the simple computation and storage requirement [8]. In addition, the major disadvantage of derivative features in background removal is the additional difficulty when dark Moving Area Filter to Detect Objects in Video Sequences from Moving Platform Sallama Athab, Hala Bahjat I World Academy of Science, Engineering and Technology International Journal of Computer and Information Engineering Vol:7, No:9, 2013 1184 International Scholarly and Scientific Research & Innovation 7(9) 2013 scholar.waset.org/1307-6892/16625 International Science Index, Computer and Information Engineering Vol:7, No:9, 2013 waset.org/Publication/16625