薄钢板搭接焊缝的微小特征实时识别  

Real-time tiny feature recognition in lap welding process

作　　者：陈海永[1] 杜晓琳[1] 董砚[1] 

机构地区：[1]河北工业大学控制科学与工程学院,天津300130

出　　处：《焊接技术》2014年第10期15-19,27,共5页Welding Technology

基　　金：国家自然科学基金资助项目(61203275);河北省自然科学基金项目(F2014202071;F2013202101);河北省高等学校科学技术研究项目(YQ2013036)

摘　　要：针对薄钢板搭接焊缝,提出了一种基于结构光视觉的图像特征识别方法。利用时空相关性原理和图像形态学开运算,消除部分随机噪声,改善了图像质量。利用峰值检测法,确定了感兴趣区域,提高了计算效率。进而利用列搜索和距离差分法实现了结构光带的特征点提取,试验结果达到了期望的性能。According to the process of lap welding for thin sheet steel, a new vision method based on structured light image feature recognition was proposed. Principles of space-time correlation of image and morphological opening operation was used to eliminate part of the random noise and the image quality was improved. Then peak detection method was used to determine the region of interest and improve computational efficiency. And use the searching method column by column and distance difference method to achieve the extraction of feature point with structured light, and the experimental results reached the desired performance.

关 键 词：搭接 机器视觉 特征提取 激光结构光 

分 类 号：TG441.7[金属学及工艺—焊接]