焊接机器人焊缝完整程度图像识别算法研究

Research on image recognition algorithm of weld seam integrity of welding robot

焊接机器人存留的焊缝缺陷会给生产企业造成安全威胁,为了让机器人有效识别焊缝缺陷,设计了一种焊接机器人焊缝完整程度图像识别方法。该方法根据采集的焊接机器人焊缝图像,采用暗通道先验理论对采集结果实施图像增强处理操作,从而达到提升图像清晰度的目的;以此为基础采用Contourlet变换将图像分解为低频分量及高频分量图像,并采用KPCA方法提取焊缝高频分量图像纹理特征,达到提高图像完整度识别效果的目的;最终将提取结果输入到构建的支持向量机模型内,结合典型缺陷焊缝纹理特征,以此识别出焊缝的完整度,完成了对焊接机器人焊缝完整程度的全面识别。试验结果表明:通过对该方法开展了焊接机器人对焊缝的裂纹、小孔、夹渣、未焊透等完整度识别测试及识别性能测试,验证了该方法的有效性强、可行性高。

The weld defects left by the welding robot would pose a threat to the safety of the production enterprise.In order to enable the robot to effectively identify the weld defects,an image recognition method for the weld integrity of the welding robot was designed.According to the acquired welding robot weld images,the method adopted the dark channel prior theory to enhance the image processing of the acquired results,so as to improve the image clarity.On this basis,Contourlet transform was used to decompose the image into low-frequency component and high-frequency component images,and KPCA method was used to extract the texture features of high-frequency component image of weld seam,so as to improve the image integrity recognition effect.Finally,the extraction results were input into the constructed support vector machine model,and combined with the texture features of typical defective weld,the integrity of the weld recognized,and the overall recognition of the weld seam integrity of the welding robot was completed.The experimental results showed that the method was effective and feasible by carrying out the integrity identification tests of the welding robot on the crack,holes,slag inclusions,incomplete penetration and other aspects of the weld and the identification performance tests.