Design Of Embedded Linear Array Machine Vision System

Surface inspection of production is an inevitable procedure of the modern manufacturing to ensure the product quality. The efficiency and precision is not high enough to meet the requirements of modern industry by traditional manual inspection. In recent years the rapidly developed machine vision technology is suitable to solve this problem. The inspection method based on machine vision can greatly increase production efficiency and automation level, as a result, the machine vision technology being the extensive researched. This thesis designed a kind of embedded linear array machine vision system to replace the traditional PC-based machine vision system, it could reduce the system cost and improve the reliability and the performance of real-time. It is suitable for industrial applications for its features of easily installed and moved.This thesis intrduced the key to building a machine vision system-CCD image sensor at first, then analysis ed the principle of linear CCD and the techinique of correlated double sampling. It discussed in detail how to building embedded linear array machine vision system, including image grabbing, image processing and network communication. This system designed the image grabbing circuit with high-speed linear CCD image sensor IL-P1-2048B, it could get high-precision image in high-speed. It used the high performance DSP TMS320C6713as core processor for image processing. Communicating between PC system with Gigabit Ethernet. This thesis introduced the system’s hardware circuit and driver software, and analysised the results of image grabbing. Finally, it introduced the embedded linear array machine vision system to the field of surface inspection. The surface inspection system contains illumination, lens, transmission and other parts.The embedded linear array machine vision system has a wide range of applications in industrial inspection. This paper aims to building a common platform of machine vision system to load image processing algorithms, and make it has a high versatility, so it can be applied to widely field of detection.