基于机器视觉的齿轮检测与测量系统的研究

发布时间：2018-08-28 10:41

【摘要】：齿轮产品在很多领域都扮演重要角色,所以必须对产品进行严格的检测。人工检测的方法存在误差大、速度慢、检测数据不能实时存储等缺点,不适合生产过程中的实时在线检测。在机器视觉软件HALCON的HDevelop开发环境下,本文设计了一种齿轮自动检测与测量的系统,包括图像采集、图像处理、图像识别等部分,主要内容为:1.将待检测的齿轮置于双轨道环形流水线的传送带上,当齿轮在传送带上运动到暗箱中的光电传感器位置时,获取齿轮图像然后将其传入计算机内进行处理。2.利用机器视觉软件HALCON对获得的待检测图像执行预处理操作。首先把获取到的彩色齿轮图像经过灰度化转换转变成灰度图像。然后,对灰度图像进行各项异性扩散滤波处理,该处理过程具有除去图像噪声的同时保留并锐化图像边缘的优点。3.齿轮检测过程中,机器视觉系统中的暗箱可以减少外界环境光照对系统的影响,所以对平滑处理后的图像采用速度最快的阈值分割算法。然后采用形态学处理和减法处理相结合的方法得到齿轮齿的个数和单个齿的面积,进而剔除不合格产品。4.在选取感兴趣区域后,结合Canny算子和双线性插值法检测齿轮的亚像素边缘。不同类型的齿轮识别过程中,提出了使用模板匹配与图像金字塔搜索法相结合的方法,以亚像素级别的齿轮中心孔轮廓作为形状匹配的模板,并且该模板支持各向异性缩放。模板匹配之后,为了能使匹配结果显示出来,对模板图像进行了一个仿射变换处理。实验证明,该处理过程的方法能快速准确的对不同类型的齿轮进行分类识别。5.获得亚像素边缘后,对亚像素边缘的轮廓应用格林定理得到齿轮的面积和中心。接着用基于Tukey的最小二乘法拟合圆形曲线,进而可以获得各个圆的半径长度。然后用一维圆弧测量法获得齿轮的齿厚、齿槽宽度以及齿距,最后经过系统标定完成测量工作。
[Abstract]:Gear products play an important role in many fields, so it is necessary to carry out rigorous testing of the products. The manual testing method has some shortcomings, such as large error, slow speed, and the detection data can not be stored in real time. It is not suitable for real-time on-line testing in the production process. The system includes image acquisition, image processing and image recognition. The main contents are as follows: 1. Put the gear to be detected on the conveyor belt of two-track annular pipeline. When the gear moves on the conveyor belt to the position of photoelectric sensor in the dark box, the image of the gear is acquired and transmitted to the computer. 2. The machine vision software HALCON is used to pre-process the acquired image. First, the acquired color gear image is transformed into gray image by gray-scale transformation. Then, the gray image is processed by anisotropic diffusion filtering, which can remove the image noise while retaining and sharpening the image. 3. In the process of gear detection, the dark box in the machine vision system can reduce the influence of the external environment illumination on the system, so the smoothed image is segmented by the fastest threshold algorithm. The sub-pixel edge of the gear is detected by Canny operator and bilinear interpolation after selecting the region of interest. In the process of identifying different types of gears, a method combining template matching and image pyramid search is proposed, in which the profile of the center hole of the gear at sub-pixel level is taken as the shape. After template matching, an affine transformation is performed to make the matching result show. Experiments show that this method can classify different types of gears quickly and accurately. 5. After obtaining the sub-pixel edge, the sub-pixel is classified. The area and center of the gear are obtained by Green's theorem. Then, the circle curve is fitted by the least square method based on Tukey, and the radius length of each circle is obtained.
【学位授予单位】：聊城大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TG86

【参考文献】