| With the continuous development of wireless communication technology and computer hardware capabilities, mobile phones, pad and other mobile devices has entered our daily life. Mobile Internet is one of the hottest topics in the field of information technology that opens a new information era. It stimulates and creates new business forms. As a result, more and more typical works which are used to be performed on personal computers are gradually migrating to the mobile terminals.Vision is the main source of information obtaining for human beings. Image, video and other visual information carrier is also one of the largest data sources of the contemporary era of big data. Computer vision has important applications in various fields, including computer engineering, communications, biology, medicine, military and so on. Video Object Tracking is a core computer vision technology. Recently, with the popularization of intelligent terminal and widely used computer vision technologies, they have been combined to create new applications. As an illustration, mobile apps, such as online video, BeautyShot, BeautyPlus,etc. So, it’s a challenging task to track moving targets accurately and robustly on the Android platform due to limited hardware resources. Therefore, this paper presents Android-based detection and tracking goals, and the corresponding algorithm has been improved.In this paper, different methods for video object detecting and tracking have been studied. Several typical object detecting and tracking algorithms have been implemented on the Android platform. The main contents of this work includes the following steps:(1) Detection target is the process of positioning object area in the image. After this step, only the foreground objects are located and regardless their classification information. Background modeling by Mixture Gaussian Model is adopted in this paper. It detects targets by modeling background.(2) When the moving objects are detected, they can be tracked by a tracking algorithm. In this article, the CamShift algorithm is adopted.(3) Experimental Analysis via the PC side for target detection and tracking. The program is wrote using C++programming language with the Android development environment and the OpenCV for android. Java is selected to write the upper UI. The JNI technology is used to call the target detection and tracking of C++ algorithms. On the Android platform, we use an Android mobile phone comes with rear camera to capture frames for human face detection and tracking experiment. The same mobile device and the software platform are also used to verify the function and performance of the proposed system. |
