پديد آورنده :
رزاقي، پروين
عنوان :
دخالت اطلاعات سطح بالا در درك صحنه
گرايش تحصيلي :
كامپيوتر-مهندسي كامپيوتر
محل تحصيل :
اصفهان: دانشگاه صنعتي اصفهان، دانشكده برق و كامپيوتر
صفحه شمار :
نه،154ص.: مصور،جدول،نمودار
يادداشت :
ص.ع.به فارسي و انگليسي
استاد راهنما :
شادرخ سماوي
توصيفگر ها :
شناسايي شي مبتني بر پاره , بخش بندي معنايي غير پارامتريك , تغيير اندازه ي تصوير بر اساس محتوا
تاريخ نمايه سازي :
10/9/93
استاد داور :
شهره كسائي،محمدرضا احمدزاده،اميرحسين منجي
دانشكده :
مهندسي برق و كامپيوتر
كد ايرانداك :
ID690 دكتري
چكيده انگليسي :
۵۵ Corporation of High Level information in Scene Understanding Parvin Razzaghi p razzaghi@ec iut ac ir October 6 2014 Department of electrical and computer engineeringIsfahan University of Technology 84156 83111 Isfahan Iran Supervisor Shadrokh Samavi samavi96@cc iut ac irAdvisor Saeed Sadri sadri@cc iut ac irDepartment Graduate Coordinator MohammadAli Khosravifard Electrical and Computer Engineering department Isfahan University of Technology Isfahan Iran AbstractIn this dissertation scene understanding is investigated The main goal in sceneunderstanding is to build a machine such that it can perceive like human and understandthe major parts of the image Scene understanding includes two important tasks objectdetection and semantic segmentation It is shown that many state of the art approaches inobject detection and semantic segmentation focus on incorporating the high levelinformation in an effective way Hence this dissertation concentrates on finding aneffective way to incorporate high level information To do this we benefit from humanthinking Hence high level information is extracted through explicit grouping of low levelinformation In many previous research works the high level information is extractedimplicitly such that it is discriminative in the entire dataset Whereas if it is obtained basedon one image and then it is completed using other images then we have betterperformance Human does it in the same way We investigated this idea in both objectdetection and in semantic segmentation In the proposed object detection method a set of discriminative parts are extracted for eachobject category through explicitly grouping of low level features In our approach in thetraining phase the object model is learned incrementally In semantic segmentation a newnonparametric approach is proposed which does not require a learning model Also regions in test image are grouped to form one semantically meaningful unit All introducednonparametric approaches are based on patch correspondence Our proposed method doesnot require explicit patch matching which makes it relatively fast and effective Also the application of semantic segmentation in content aware image retargeting isinvestigated In image retargeting each human based on his understanding of image produces a different retargeted image This is due to that different semantic classes havedifferent degrees of importance for each person Hence in our approach a priority level isassigned to each semantic class Key WordsScene understanding high level information object detection semantic segmentation image retargeting 1 IntroductionObject recognition is one of the most important research areas in computer vision Up tonow many approaches are introduced in the field of object recognition Some of theseapproaches are part based in which parts of object are placed in a special structure In
استاد راهنما :
شادرخ سماوي
استاد داور :
شهره كسائي،محمدرضا احمدزاده،اميرحسين منجي
