DSP-based Narrow-gap Submerged Arc Weld Image Processing And Feature Extraction

Automatic seam tracking technology,especially the visual tracking, has become a key technology for domestic and international welding automation.It is more attention, and to focus research. Weld image processing technology is the core of seam tracking system. The most basic image processing technology is to capture and store the image. In the welding process, the need for real time image acquisition and storage. To improve image processing speed, high-performance DSP processing equipment used. BlackfinDSP561has a powerful computing capabilities. It has a dual-core architecture that can simultaneously image processing and storage tasks. It’s Visual DSP++programming platform using intuitive, simple user interface, allows users to perform various operations flexibility. Visual DSP++supports C/C++compiler. It is beneficial for post-program improvements, improving the development efficiency.In the narrow gap submerged arc welding, the welding flux was completely buried. Can not know whether the gun position in the weld centerline. Issued a linear light source light,at the forefront of flux reactor. Image sensor with real-time image data acquisition welding position. Completed ahead of torch position to determine the deviation. The position error signal to the actuator controller. The position error signal to the actuator controller to achieve the automatic seam tracking.Based on narrow-gap submerged arc weld image features, design the appropriate image processing steps. Image enhancement, making the image more clear weld. Image segmentation, the weldment and weld the edges of the split. Image filtering, image segmentation, after eliminating the weld near the edge of a small noise. Edge detection, to be clear and complete picture of the weld edge. Narrow-gap submerged arc weld track, mostly straight, curved. Specifically proposed a modified Hough transform to extract the edge of the weld. Specifically proposed a modified Hough transform to extract the edge of the weld. Compared to the traditional Poberts, Prewitt edge detection operators such as, Hough transform detection of weld image more clear and complete. While weld image processing, the collected data is compressed. Designed for narrow-gap submerged-arc welding method of image compression. Ensure that weld characteristic value is not destroyed, reducing the video storage space. Save weld image, which will help to track the quality of the weld compared. Data analysis for the latter part of the original data.