Gpu-based Design And Realization Of Real Time Imaging Of The Weld

In industrial inspection process, the weld containing pores, slag, cracks and other defects will affect the product quality and safety, so the weld defect detection in the image is extremely important. Currently, requiring large inspection workload, high sensitivity, and rapid and intuitive test results, traditional techniques can not achieve real-time processing, although we use variety technologies in the detection of the boiler tube weld defect, However, in recent years, the rapid development of programmable graphics processor GPU improves the speed of computer graphics and its related applications. Therefore, using GPU-based real-time image detection system can complete line detection and analysis, image processing can accelerate the weld speed and improve the image detection accuracy.In this real-time processing for the image of the weld process, The image processing speed is slow,X-ray image clarity is poor with low resolution, which is not conducive to welding image analysis and detection. To overcome these problems, the paper presents GPU-accelerated image processing methods used by programmable graphics processor to improve image processing speed and image clarity. This method uses the GPU parallel processing architecture, making full use of its powerful computing capabilities and high parallel computing speed, improving system efficiency, shortening the processing time for each frame to achieve a dynamic real-time image processing.In the overall design process, firstly, image of the weld process detailed investigation and analysis and understood the weld image processing processes, methods and problems and of the related techniques. Secondly, select the appropriate design algorithms and techniques, through the DirectX interface technology to build a GPU-based processing platform. Finally, by programming to improve the image processing speed, clarity and resolution, to achieve bulk welding process image processing functions fast real-time requirements of the industrial process and use of a high reference value.