【摘要】 一般的拼缝检测视觉传感器系统传感器和焊炬之间存在前视距离,会无法避免的造成跟踪误差.因此,本文提出了一种无前视距离的被动光视觉传感器测量窄间隙拼缝,并使用视觉传感器直接观测熔池,避免了前视距离造成的误差.同时,一般的视觉传感器同一时刻只能采集一个拼缝点的图像,图像处理时极易受到光噪声的影响.本文采用加装远心镜头的高动态范围的相机,同时利用外部触发方式减小曝光时间,可获取含显著拼缝特征的图像.相应的,提出了一种改进的霍夫变换方法用以实时拼缝提取.该方法对霍夫变换图像处理算法的搜索区域、搜索模板和权值计算方法进行了优化,兼顾了霍夫变换的鲁棒性同时保证了算法效率,能准确提取和追踪拼缝的斜率和截距.更多还原

【Abstract】 Vision sensor system based on structured light has been widely used for seam detection. However,there still exist preview distance between vision sensor and welding torch in general measurement system,which can inevitable cause seam detection errors.This paper proposed a passive vision sensor without preview distance to measure narrow butt joint. The weld pool is observed directly by vision sensor to avoid the error caused by preview distance. Meanwhile,only one point of the weld seam can be captured at a time,which makes the image processing vulnerable to light noise. So the high-dynamic-range camera is used and adjusted to a specific position to obtain image containing significant feature of weld seam. Accordingly,an image processing method based on improved Hough transform is proposed to extract the seam position. The search erea,search template and weight calculation method of Hough transform are optimized,so the method can guarantee the robustness and efficiency and accurately extract the seam position and track the slope and intercept of weld seam.更多还原