The Research And Implementation Of The Platform Of 2D Image Acquisition Based On Stereo Vision

With the rapid development of semiconductor industry and digital image processing, machine vision have been applied to many fields,from daily living to aviation technology.So,there will be a bright prospect for machine vision.Three-dimensional reconstruction is the sophisticated application based on machine vision, which is also called binocular stereo vision. In this system, two cameras are used as the collecting devices of the image data to measure the object. The platform of traditional stereo vision generally consist of computer's board-card. Nevertheless, such methods is not quite flexible, and the price is very high. Therefore, the system, in the paper, will consist of FPGA and ARM, and choose TCP/IP as main communication method with PC. Then, which could satisfy the demand of multi-cameras co-working and the data transmission of long distant.In the system, image compress and pre-processing chip is a piece of FPGA (XC3S500E) of Spartan3E series provided by Xilinx company, which has 5000,000 system gates. STR912 based on ARM9 core is main control chip in the device, which will perform some simple missions of control and data transmission through TCP/IP protocol. FPGA directly control the image sensor, and data gained from the image sensor is stored in SRAM. Then, ARM9 will read data from SRAM, and transmit to PC. Finally, PC will realize the algorithm of image processing.In the ending of paper, in order to validate the theory of 3D reconstruction, we established a device of binocular stereo vision,and measured the 3D data of the object.