Research On Motion Object Detection And Tracking Methods Based On Complex Background

Moving object detection and tracking algorithm has become one of the important researchprojects in the current image processing and computer vision fields, for its widely applicationsin the fields of military guidance, safety monitoring, intelligent traffic and others. Because ofthe influences from the factors including camera position and angle changes, illuminationchanges, objects’shadows in the background and so on, research on moving object detectionand tracking algorithm in the complex background, has become one of the research hotspotsnowadays.On the basis of the analysis and summary of existing algorithms which are commonlyused, this thesis mainly aims person and vehicles to investigate the moving object detectionand tracking algorithms based on complex background. First preprocess the obtained originalimages, then detect if there are moving objects in the images which are after preprocessing,extract the object area images and postprocess them. Use moving object tracking algorithm totrack the moving objects, and lock them finally. The main research work can be summarizedas follows:(1) On the image preprocessing, based on the test and analysis of existing imagesmoothing and image sharpening algorithms in commonly used, choose the median filteringalgorithm and Laplacian mask sharpening algorithm to do the image preprocessing operations,and improve median filtering algorithm for its defect of slow processing speed. The operationcan remove noise in the image effectively, highlight the edges of the moving object, andfacilitate the subsequent processing such as moving object detection, etc.(2) On the moving object detection, main concern is the moving object detectionalgorithm based on image registration and consecutive four frames difference multiplication.Use the algorithm to analyze the images which are after preprocessing, and successfully detect and obtain object area images in the complex background. Then do the postprocessingoperations to the object area images including morphological processing and connectedcomponent, successfully remove interference informations in the object images, and obtainthe positions of the moving objects in the original images. This operation provides the basisfor tracking the moving objects correctly.(3) On the moving object tracking, main concern is the object tracking principles andrealization processes of Mean Shift algorithms based on color histogram features and varianceratio features respectively. Then test and compare the tracking performances of twoalgorithms through analyzing the object result images obtained in the detection process, andchoose the Mean Shift object tracking algorithm based on the variance ratio features finally. Itcan track the moving objects in the complex background timely and accurately.Under VC++6.0 platform, by using C language combined with some function fromOpenCV library, realize the above algorithms on the image preprocessing, moving objectdetection and tracking. Verify the effectiveness of the algorithms through the tests, and finallyrealize the moving object detection and tracking based on the complex background.