Research On Moving Target Detection And Recognition For Mobile Phone Videos

With the rapid development of Internet and multimedia technologies, video image processing and computer vision have gradually been widely applied. Researchers have paid much attention to target detection because of its great economic value. Recently, as smart phones are getting more and more popular among young people, the photography and video function is one of the most important and popular features in the smart phone. Refining the performance of smart phone’s photography function is a significant way to enhance user experience on smart phone.In this paper, traditional moving target detection and tracking and characteristics of mobile phone shooting environment are deeply studied. We improve Camshift and propose a target detecting algorithm based on background compensation and three frame difference method. The new algorithm proves to be more accurate and efficient. In this paper, our work is mainly focused on:(1) Basic knowledge of image pre-processing, image enhancement and other digital image processing are studied and analyzed. Taking into account the scenarios for mobile phone photographing, video sequences are taken to do efficacy testing(2) The algorithms for target detection under static and dynamic background are studied. Considering the weakness of traditional target detection algorithm under dynamic background and the special nature of mobile shooting, we propose a new moving target detecting algorithm based on background compensation and three frame difference to complete the multi-target detection for real-time mobile video. The accuracy and efficiency of the algorithm are verification by experiments.(3) Classical moving target tracking algorithms such as Camshift is studied, this paper refines Camshift by using LBP feature. The new algorithm is proved to be accurate and efficient by experiments.(4) For the mobile phone shooting, this paper combines target detection module and tracking module. The combined algorithm is accurate and efficient under the testing videos, which creates a foundation for auto focusing and other applications.