The Design And Implement Of Real Time Sight Positioning System Based On FPGA

Sight-Positioning is an emerging research topic, which has immeasurable development potential both in terms of human-computer interaction field and behavior analysis aspect. In this paper, after referring to a variety of sight detection means and actual demand, the method based on image processing technology is finally adopted to design and realize the real time sight positioning system. Specifically, the main work includes the following several aspects:(1) Research status of sight detection is introduced in detail, and the process of development and the basic methods of sight detection are discussed. Furthermore, its advantages and disadvantages have been analysed.(2) The development platform of this system is briefly discussed. Mainly includes ZYNQ’s startup and configuration process, Linux system transplantation and QT library cross-compiling.(3) The overall framework and specific function module design of this paper are put forward, and the key technologies in this design are elaborated.(4) Analysis and study have been made on the sight detection technology based on image processing method, and the necessary improvements on algorithms are made, such as the improvement of sorting algorithm for median filter function which would save 30% of the time in the worst case after the improvement; the projection integral algorithm is efficient for face detection, which greatly reduces the complexity of the algorithm; the introduction of coarse positioning algorithm of narrowing eyes search area, which greatly improves the detection accuracy of eye window. All these measures have ensured the real-time and accuracy of the system.(5) With the real-time and accuracy as the evaluation criteria, a full range of tests on sight positioning system have been done. And analysis of the main problems existing in the system has been done, and the solution has been put forward. The test results show that, the system can basically meet the design requirements in both real-time and accuracy.The system uses ZedBoard development board as the hardware platform, and adopted the hardware and software collaborative design approach. In terms of hardware, ZYNQ dual core Cortex-A9 ARM is used as the control core to complete the running of Linux system and the operation of application software interface. The system takes FPGA as the critical device for image processing algorithm. In terms of software, embedded system configuration and code writing and debugging are mainly completed with ISE development tools provided by Xilinx Company, and application development in QT provides a visual man-machine interactive interface.