基于双目视觉的工件测量研究  

Research on workpiece measurement based on binocular vision

作　　者：潘琪[1] 秦襄培[1] 尹雄[1] 武胜超 曾诚[1] 

机构地区：[1]武汉工程大学机电工程学院,武汉430205

出　　处：《智能计算机与应用》2018年第1期18-21,26,共5页Intelligent Computer and Applications

摘　　要：采用双目视觉代替人眼对工件进行测量提高了生产的柔性和自动化程度。双目视觉的精度很大一部分取决于图像的预处理过程,Canny算法作为传统的边缘检测方法在边缘定位和抗噪声干扰性达到了较好的折中效果。本文提出了一种先进行Sobel算法进行边缘检测,后采取形态学运算进行细化的方法进行双目视觉的测量,将2种方法对比证明了该方法在精度上能达到Canny算法的检测精度,且不会出现边界丢失现象。Using binocular vision instead of the human eye to measure the workpiece can increases the flexibility and automation ofproduction. The accuracy of binocular vision depends in part on the image preprocessing, Canny algorithm, as a traditional edgedetection method, achieves a better compromise in the edge of the positioning and anti-noise interference. The paper proposes amethod that uses Sobel algorithm in edge detection, and then implements morphological operations to refine the measurement ofbinocular vision. A conclusion drawed from comparing the two approaches is that both of them can reach the accuracy of Cannyalgorithm without any boundary loss.

关 键 词：双目视觉 CANNY算法 SOBEL算法 形态学运算 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]