Study On Multiple Image Processing Hardware System Based On DSP

The panoramic vision system and the stereo vision system are widely used in various fields because of their special function being different from the ordinary vision systems.But because of their disadvantages as panoramic vision losing of depth information and stereo vision generally having a narrow field of view, there are many limitations in using them. Panoramic stereo sphere vision theory proposed by our research group is advanced in the integration and optimization of these two systems. It can obtain the panoramic image and can extend the field of stereo vision,and a reliable hardware system is the key part to prove the feasibility of this theory.With the development of embedded technology in the application and development of the field of image processing, there are various efficient image processing schemes. This paper adopts embedded multi-channel image processing hardware system based on DSP to realize the image data collection, caching and output, at the same time to meet the hardware processing requirements of algorithms. The system uses DSP+FPGA structure, with using DSP as the system scheduling core and complex algorithm processing core, using FPGA as the core of image preprocessing in management of multi-channel image acquisition logic and image preprocessing because of its powerful logic management capability and hardware processing high-speed characteristics, using CMOS image sensor to collect source images.Because of the source image by embedded hardware system can not be obtained ideally,it must also be fixed on the packaging system. Through researching and designing a mechanical structure, each part of this system are fixed in special position of space, to assist the whole system realization, and to get source image used to panorama stitching and stereo matching.This paper designs and realizes the schematic circuit diagram and the printed circuit board of an embedded hardware system, completed the construction of the internal logic of image preprocessing module, through the three-dimensional simulation determined and made the system mechanical structure. Finally, assembling the whole system to debug and test. The experimental results show that the system can realize the multi-channel image acquisition,caching and output, to prove the feasibility of the proposed scheme but also reveals some shortcomings.