论文部分内容阅读
提出了一种基于灰度形态学的图像处理算法,有效去除了线结构光视觉传感焊缝跟踪图像中飞溅、弧光和烟尘等干扰。根据飞溅在图像中表现为明暗相间的条纹的特点,采用灰度形态学开操作,通过合理设计结构元素,去除了飞溅干扰。针对不同图像受弧光、烟尘影响不同,背景灰度值差异较大,不利于二值化阈值选取的问题,采用图像相减和顶帽变换消除其影响。最后,采用动态开小窗与最小二乘法相结合的方法准确提取到了焊缝特征点。该算法能有效去除飞溅、烟尘等干扰,实时性好。
An image processing algorithm based on gray morphology is proposed, which effectively removes the interference of spatter, arc and soot in the tracking image of the line-structure optical vision sensor. According to the spatter in the image shows the characteristics of light and dark stripes, the use of gray morphology to open the operation, through the rational design of structural elements, to remove the splash interference. Different images are affected by arc and soot, and the gray value of the background is greatly different, which is unfavorable to the threshold selection of the binarization. The image subtraction and top cap transformation are used to eliminate the influence. Finally, the characteristic points of the weld are accurately extracted by the combination of dynamic open window and least square method. The algorithm can effectively remove the splash, smoke and other interference, real-time good.