Research On Some Key Technology Of High-speed Image Processing Plantform In Machine Vision

The function of machine vision is to simulate human vision using machines in order to detect and measure in the images. Machine vision is an effective way to achieve intelligent and automated equipment with precise control. Many tasks which human can not handle could be completed with the help of machine vision, while helping to improve product quality and production efficiency. According to authoritative international statistics, the total global machine vision market is about 90 billion dollars with an annual rate of about 10.9% of rapid growth in 2007. More than 15 billion could be achieved in 2012. It shows that machine vision system is of great significance and broad market prospects. The future direction of development of machine vision systems is the high-speed, integration and personalization, so an embedded high-speed image acquisition and processing platform with general-purpose is needed to meet the above needs.Currently the domestic machine vision manufacturers are mainly the agents for foreign products, including the hardware products, especially the high-speed and high-resolution hardware for high-end applications. They are also aware of the needs of developing their own products, and some research institutes have studied the image processing hardware, but mainly in the low-end applications.An embedded general high-speed image acquisition and processing platform (EGHIAPP) based on FPGA and DSP is designed in this thesis based on the research on the work of high-speed image acquisition and processing in the machine vision system. The critical issues of high-speed image acquisition, compression and storage are studied. The main works are as follows:1. The functions which the high-speed image processing platform should possess in the future applications are proposed according to the future direction of the development of machine vision, on which EGHIAPP is designed.2. The image acquisition problems are researched on the Camera Link and GigE cameras in the EGHIAPP. The detailed engineering realizations of the image data acquisition are designed towards the two types of camera interface. Finally, the high-speed image acquisitions of two typical cameras are implemented respectively.3. The issue of high compression speed problem as well as the problem brought by the large capacity buffer in the traditional scheme are addressed based on the research of the high-speed JPEG compression of the real-time collecting images. The real-time JPEG compression tasks of color images with large resolution of 5 million and gray images with the high-speed of 500 frames per second are realized using only a single FPGA chip.4. The data read and write capabilities of SATA hard disks using FPGA are realized on the research of the mass storage problem of real-time images. The causes which impact the speed of the hard disk are analyzed in order to enhance the depth of the hard disk storage. The parallel disk storage technology is also researched. Finally, the parallel storage solution is given in the example of two-channel disk storage.5. Two typical applications in the machine vision areas are completed with the help of the above designed EGHIAPP, which illustrates the good generality feature of the platform. And one of the applications has been successfully applied in real production, and achieved good economic and social benefits.