基于图像处理的微小塑料齿轮轮廓优化
摘要:通过计算机视觉系统的构成、图像预处理、图像分割和数字图像的像索连通性理论与技术的研究提
出,微小塑料齿轮的二值图像噪声点去除及齿轮轮廓提取方法,用数学形态学方法提出轮廓的简化算法采用边缘滤波器对图像进行去噪处理。用Robot边缘检测算法进行边缘提取齿轮的轮廓,统计出大齿轮和小齿轮上面齿的个数和每个齿轮上齿之间的角度。实验结果表明, 该优化轮廓的简化算法, 可获得准确的齿形检测数据, 能满足工程测量的实际需要。
关键词:塑料齿轮 二值化 边缘处理 roberts算子
Contour Optimization of the Micro Plastic Gears Based on Computer Vision
Abstract:On the basis of the studies upon the computer vision system construction and the theories and technologies of the image preprocessing, the segmentation of image as well as the pixels connectivity of digital image, the methods of wiping off the noise points in the binary image of the micro plastic gears and extracting the contour of the gears were put forward. The simplifying algorithm for the contour extracting was raised by the methodology of the mathematical morphology. T he principles and the implement of the key technologies of the above algorithm were also presented. It found out that a curvilinear could be represented with the fewest points by deleting the unassociated points. At last, the experiments have showed that the inspection dates of the tooth profile could be obtained accurately by the simplifying algorithm of contour optimization and the algorithm could meet the needs of the virtual engineering inspection.
Key words:plastic gear binarization Edge processing Roberts operator
正文:
引言:
塑料齿轮的齿崩、缺齿、披锋、翘曲变形等缺陷[1]容易导致齿轮传动的噪声、磨损加剧、效率降低甚至传动系统的卡死现象。微小塑料齿轮缺陷产生的种类、大小、程度与分布位置都是随机的。因此, 用接触式检测不仅难度大, 而且效率低。机器视觉技术是非常有效的非接触检测技术, 被广泛地应用于各种加工件的在线检测和高精度、高速度的检测技术领域, 因此采用 CCD 图像测量和识别技术做非接触检测, 是解决这一问题的有效途径。本文主要研究采用计算机视觉技术, 检测微小塑料齿轮时轮廓的提取和优化方法。
具体内容:传统的算法主要根据计算机视觉原理和二值化处理去除:具体原理如下, 微小齿轮计算机视觉检测系统由CCD 传感器、阈值与最大类间方差区域切割综合法, 较好
光学系统、计算机数据采集和处理系统输出控制等部分组成。根据所检测齿轮的特点并结合现有的实验条件,选用了BASL ER A102f CCD 数字摄像头和相配套的IEEE1394 图像采集卡、TEC2M55 焦阑镜头图像识别处理流程如成像系统获取的原始图像由于受到种种条件限制和随机干扰, 往往不能在视觉系统中直接使用, 必须在视觉信息处理的早期阶段对原始图像进行噪声过滤等图像预处理,有一定的
局限性。 本文采用边缘滤波器对图像进行去噪处理,与均值滤波器和中值滤波器相比, 边缘保持滤波器既能滤除噪声,又能很好地保留塑料齿轮图像中的轮齿齿廓细节。图像分割是把图像分成各具有特性的区域并提取出感兴趣目标的技术和过程。针对微小塑料齿轮图像的特点, 本文采用迭代
解决了图像多阈值分割的问题。计算机视觉检