基于机器视觉的数控铣床考工零件检测系统设计

Design of Parts Detecting System for NC Machining Skills Examinations Based on Machine Vision

传统数控技能考试的零件检测均为人工检测,存在一定的主观因素,判定的结果有偏差,而且效率较低。为了考核的公平并且提高检测效率,构建基于机器视觉的数控考工零件检测系统。系统通过摄像头采集被测工件图像数据,并对图像进行中值滤波、灰度拉伸、二值化、边缘检测等预处理;应用基于Sobel算子的改进算法,获得了较好的边缘提取效果。采用径向搜索方法完成对边缘像素的识别,通过计算与标称中心点的距离获得工件测量值。为了使测量结果具有比较性,引入方差分析,对测量样本进行均值计算,同时计算出方差,得出置信度,对检测结果偏差进行描述。通过与人工检测相比:采用该系统,检测效率提高了3～5倍,准确率提高了10%以上。

Workpiece inspection in traditional NC skills examination is tested by manual. Since there are certain subjective fac- tors, the results have deviation and low efficiency. In order to improve test efficiency and fairness, the workparts inspection system for NC skills examination was constructed based on machine vision. In this system, the workpiece image data were measured by camera. Pretreatment operations for the image, such as median filtering, gray-level stretch, binarization, edge detection and so on, were made. By using an improved algorithm based on Sobel operator, the good edge extraction effect was obtained. The edges pixels identification were completed by using radial search method, and the workpiece measuring value was obtained by calculating the distance to nominal center. In order to make the measurement result comparative, variance analysis was introduced. The mean value calculation was made to measuring samples, the square variance was calculated simultaneously, confidence was obtained and testing results deviation was de- scribed. Comparing with manual detection, the efficiency is increased 3 -5 times, and accuracy is increased by 10% or more by using this system.