| In the seam automatic tracking technology, welding-line image processing system is a key part. Its function is process image collected by weld sensor; extract the characteristic parameters of weld pool. After comparing with scheduled welding track, it gives signal of welding-torch location and judges whether the location is correct or not. So it influences whether closed-loop control is achieved and controlling effect. It has an effect on welding quality. In order to enhance performance of welding-line image processing system, many specialists and researchers have designed lots of means, a majority of which are how to improve software and arithmetic, but in hardware, it usually runs on huge volume industrial computer.At present, embedded system has entered into SOPC technique based on Niosâ…¡soft core, whose character is small volume, high performance, low electrical consumption, design quickly and flexibly and so on, Niosâ…¡is designed specially for SOPC, configurable Niosâ…¡empowers designer to modify hardware easily as software programming, it brings a fire-new thought in embedded design.The paper's purpose is enhance performance and reduce volume of welding-line image processing system by using Niosâ…¡soft core. The paper has two parts. First part introduces structure of Niosâ…¡, it includes FPGA chips which support Niosâ…¡and components of Niosâ…¡, then introduces how to design Niosâ…¡. At last of first part, ensure how to build system based on Niosâ…¡via analyzing design flow fully. In the second part, design system of welding-line image process based on design flow. The paper introduces how to accelerate Log grey change and Laplace arithmetic with custom instructions in detail. And discuss Niosâ…¡can accelerate processing speed within hardware instrument field. At last, processed welding-line images are streamed into computer by using EZ-USB chip. |