The Development Of Real-time Video Processing Platform Based On ARM

Image processing technology is widely used in product counting and quality testingof high-speed industrial product line and other fields. The embedded devices has thecharacteristics of low power, low cost and portability, which makes it have greatadvantages and broad prospects in high-speed real-time image processing when combineswith parallel processing technology. The video enhancement is a common method used inthe field of video processing pretreatment. The real-time video enhancement realized onthe embedded platform will reduce the difficulties of development and improve the effectof detection when realize the real-time image detection combined with the FPGA.The hardware platform used for parallel visual inspection system selects the AM3359which is a product of TI and use Cortex-A8architecture as the main processor. Theschematic design are performed on various modules of the hardware platform, and thePCB layout design are performed on six layers board. The whole hardware platformincludes a processor module, power module, memory module, video interface module andother peripheral interface module. And FPGA, as the coprocessor of hardware platform,communicates with the main processor through the built-in block RAM. The research ofvideo enhancement selects the histogram equalization algorithm as a starting point, whichhas good real-time and obvious enhancement effect in existing image enhancementalgorithms. A variety of fast computing enhancement algorithm based on histogramequalization are realized and compared on the embedded platform. Additionally, theauthor suggests improvements for BIDHE and CLAHE algorithm.The author uses the host computer to simulate the operating environment ofhigh-speed production lines in which embedded platform works and completes real-timevideo enhancement system which consists of video capture module, compression-transmission module and video processing-display module. The system has been tested and achieves the expected performance. It verifies the effectiveness and feasibility of theARM platform for real-time video processing.