A deep learning-based welding spatter segmentation and identification system and method

Abstract

The application discloses a kind of based on deep learning's welding spatter segmentation and identification system and method, setting image acquisition module and image pre-processing module, utilize improved deep learning network to build spatter segmentation module, pixel-level segmentation is carried out to the image after processing, and welding spatter area is identified, and initial segmentation result is output, the initial segmentation result is carried out morphological processing and boundary optimization, and segmentation result is generated, segmentation result is converted into position information under actual coordinate system, provide guidance for automatic polishing equipment, detection result is superimposed and displayed on original image, provide intuitive detection result.The detection precision of white body welding spatter is significantly improved, and it has stronger adaptability to scattered and irregular welding spatter;It can handle detection tasks under complex background and changing light conditions;Real-time performance is good, meet the production line beat requirement;For automated polishing equipment, accurate area positioning information is provided, and polishing efficiency and quality are improved.

Images