【摘要】 针对激光传感图像研究了用于焊缝跟踪的实时图像处理算法,以典型的V型坡口焊缝为例展开论述。就中值滤波运算量大的缺点,提出了改进的中值滤波算法,省去了模板元素排序环节;采用灰度比较得到最大值的方法巧妙实现图像细化;充分利用点与直线的距离关系和激光结构光图像的特点可靠捕捉到跟踪特征点。在图像处理的关键环节上实现了高效的图像处理,试验证明这些方法可靠性高,具有抗杂散光或激光反光的独特优点,能够满足焊缝跟踪系统实时性的要求。更多还原

【Abstract】 The real-time image processing algorithms for laser sensing was especially studied in this paper,including median filtering,the image thinning,feather point extracting etc.The following discussions are mainly based on the common V groove welding seam.In terms of too much operations of median filtering,an improved median filtering algorithm was put forward,so the traditional ordering algorithm for template elements was not needed any more.In image thinning,the row and column of the pixels with max gray value were considered as the thinning target points.According to the characteristic of the image,the relation of distance between line and point was employed to extract the feather point.These improved algorithms with excellent functions of anti-disturbance or anti-reflecting were demonstrated to be effective,simple and very feasible for laser sensing by several experiments.The real-time performance that welding tracking system requires was achieved by the above methods.更多还原