Dissertation
Dissertation > Industrial Technology > Automation technology,computer technology > Computing technology,computer technology > Computer applications > Information processing (information processing) > Pattern Recognition and devices > Image recognition device

The Research of Moving Object Tracking System Based on Embeded Image Process Unit

Author SunLianMing
Tutor ZhaoJie
School Harbin Institute of Technology
Course Mechanical and Electronic Engineering
Keywords moving object tracking object detection and location embedded System image processing
CLC TP391.41
Type Master's thesis
Year 2008
Downloads 243
Quotes 0
Download Dissertation

Moving object tracking has very important application value and broad development prospect in the fields of military, industry, scientific research, medical and so on. While image processing unit is the core of moving object tracking system, its performance directly determines tracking effect. Embedded system has great advantages in many aspects, such as real-time performance, stability, reliability, flexibility and miniaturization. With the rapid development of embedded technology, it will play a more and more important role in the field of image processing. On such background, the author develops an embedded image processing unit, and studies the moving object tracking method at the same time, and finally builds experimental system of moving object tracking.On the base of lucubrating the inland and overseas development of embedded image processing system, according to system functional requirement and modularization design rule, the author accomplishes the design of DM642 configure module, expanding memory space module, video module, network transmission module, and power module utilizing the digital signal processor TMS320DM642 as core unit. So the whole embedded image processing hardware platform is set up.Meanwhile, the author studies the method of color-based moving object tracking. In YUV color space, an improved distribution space of object color is described using Boolean vector, and an improved region-labeling algorithm based on Run Length Encoding (RLE) is put forward. The problem of run-labeling conflict is effectively resolved. What’s more, the less memory space is used. By this method, object detection and location can be realized quickly and accurately. Thereupon tracking goal can be achieved through the two links of object motion state estimation and camera control.Then the author, adopting modularization design idea, develops the whole software system of the embedded image processing unit based on DSP/BIOS real-time operating core and RF5 reference framework. Thus the main functions, such as video capture and display, object detection and location, and network transmission are realized. In the end, the author founds the experimental system of moving object tracking using related hardware. Moreover the performances of every soft hardware functional module and the whole system are validated through color image segmentation experiment, object detection and location experiment, network transceiving experiment, and comprehensive tracking experiment.

Related Dissertations
More Dissertations