پديد آورنده :
حيدري، وحيد
عنوان :
تشخيص و دسته بندي وسايل نقليه بر اساس پردازش تصوير براي يك سيستم نظارت ترافيكي
مقطع تحصيلي :
كارشناسي ارشد
محل تحصيل :
اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر
صفحه شمار :
نه،95ص.: مصور،جدول،نمودار
يادداشت :
ص.ع.به فارسي و انگليسي
استاد راهنما :
محمدرضا احمدزاده
توصيفگر ها :
بخش بندي اشياء متحرك , تنظيم دوربين , تشخيص همپوشاني وسايل نقليه , حل همپوشاني وسايل نقليه , رديابي وسايل نقليه
تاريخ نمايه سازي :
22/4/92
استاد داور :
بهزاد نظري، مازيار پالهنگ
دانشكده :
مهندسي برق و كامپيوتر
چكيده فارسي :
به فارسي و انگليسي: قابل رويت در نسخه ديجيتالي
چكيده انگليسي :
96Abstract This thesis presents a new method to classify vehicles with resolving vehicleocclusions in traffic images Moving objects are detected in an image sequence capturedfrom a traffic scene using a motion segmentation algorithm Then the foreground objectscorresponding to the occlusion of vehicles are detected and most of them are split into theoccluded vehicles Finally the divided objects and the objects corresponding tounoccluded vehicles are classified Partial occlusion of vehicles is detected by evaluatingthe convexity of the foreground objects Then partially occluded vehicles are split by theso called dividing line of the occlusion region The divided objects are classified intosmall and large vehicles by evaluating their normalized size If the object is not partiallyoccluded its normalized width and the ratio between length and width is extracted todetect if it is a full occlusion and classify it by developing a hierarchical classifier To evaluate the proposed method quantitatively we use several traffic image sequencesin which the occlusion of vehicles is common Then we compare it with state of the artmethods The accuracy of the proposed method and the comparison with state of the artmethods demonstrate that the proposed method is efficient We also implement twomotion segmentation algorithms averaging with variable threshold which is equivalentto single Gaussian method and Gaussian Mixture Model GMM Then we analyze theresults quantitatively and qualitatively and compare the methods The experiment isperformed on an image sequence captured from an urban traffic scene including scenelight changes vehicles stop and background parts movement The results show thatGaussian Mixture Model GMM has better performance and we use it to detect movingobjects for the proposed vehicle classification and occlusion resolving method Keywords Traffic monitoring system moving object segmentation camera calibration occlusiondetection occlusion resolving vehicle tracking vehicle classification
استاد راهنما :
محمدرضا احمدزاده
استاد داور :
بهزاد نظري، مازيار پالهنگ
