一种亚像素图像边缘检测算法及其在智能灌墨包装线上的应用

A sub-pixel image edge detection algorithm and its application in intelligent ink bottle packaging line

为解决墨水瓶自动灌装线上墨水瓶盖的把手边缘位置检测问题,提出一种基于二次曲线拟合的2步定位亚像素边缘检测算法。通过降维、粗精2步定位和拟合模型的可更换,减少运算时间,提高算法的适应能力,达到快速准确定位物体边缘的目的。应用亚像素边缘检测算法设计一种双探测线法并应用到墨水瓶盖的灌装线上,实现对瓶盖把手边缘的快速准确定位,把手的角度测量误差能够控制在3°以内,平均检测时间为2.15 ms,保证机器人夹取瓶盖时能够有效避开把手位置和高抓取效率。

In order to solve the problem of fast and accurate positioning of ink bottle cap handle on the ink bottle automatic filling line,a two-step sub-pixel edge detection algorithm based on quadratic curve fitting was designed.Through dimensionality reduction,coarse and fine two-step positioning method and replaceable fitting model,the computing time was reduced to improve the adaptability of the algorithm so as to achieve fast and accurate positioning of object edges.A double detection line method was designed and applied to the ink bottle filling line to achieve fast and accurate positioning of bottle cap handle edges.The measured angle error of the bottle cap handle can be controlled within 3°and the average detection time is 2.15 ms,which ensures that the robot can effectively avoid the handle position when clamping the bottle cap,and can have high grasping efficiency.