A Dimension-Measurement System Of Pipe-Hole Accessory Based On Machine Vision

Machine vision, which can improve the capability and flexibility of robot, is one of the main means from which industrial robot gets surrounding information. Applying from the specific industrial surrounding, a total production line equipped with machine vision measurement system, as well as the measurement software embedded into the industrial computer, is established. This measurement system has worked normally, with high efficiency and stability.The dimension-measurement system is expatiated following with 8 facts: Selection of Hardware, Establishment of Mechanical Platform, Calibration of the CCD System, Image DAQ, Image Preprocessing, Image Analyzing Arithmetic, Error Analyzing, and Software. A special calibration method is hold to enhance the adaptability to various of work pieces and a serial of arithmetic are compared to analyzing the precision, the validation and processing speed of the measurement.During the whole developing process of this system, the main work includes: 1 Instead of the original method of using the net-grid surface, a standard accessory be calibrated by precise instrument is used as the reference in the CCD camera calibration model.2 In the image preprocessing, it analyzes results of several kinds of smoothing filter;fast median filter stands out for its good edge maintenance, adaptability and speed. This article takes fast median filter as the way to process the collected image. Also it uses OTSU thresholding arithmetic to segment image.3 In the edge detection part, the performance between classical edge algorithm and canny algorithm is compared. Canny algorithm is better than classical edge detection algorithm in the way of the precision of edge locating and anti-noise. So the algorithm is used in the paper.4 The methods of Hough Transform and advanced Hough Transform for inspecting the circle edge are given on the basis of study of visual inspection in the paper. The method is used to detect circle by an improved algorithm that is faster and more precise than old one.