Study On Measurement Techniques Of Position Degree Of Hole Group Based On Digital Image Processing

Parts of hole group are used extensively in mechanical products. Because the quality ofthe hole- group parts can determine whether it can be assembled correctly with shaft or otherparts, both researchers and engineers have paid high attention to the error measurement for along time. Three coordinate machines and synthetic gauges are both usually used to measurethe position degree. The former is very expensive and much hard to be operated, while thelatter, as a kind of unscaled measuring tool, needs manual operation that is laborious andtime-consuming.Digital image processing technology, as a novel non-contact detection, can overcome thedisadvantages of manual operation, improve the detecting precision and efficiency, and reduceproduction cost. However, there is not a universal system of image measurement till now.Aimed at different tasks or objects, we need to establish image measurement system ofparticular applications, especially the software system for measurement.The measuring system of the position degree for hole group based on digital imageprocessing technology is studied in this dissertation. Considering the difficulty andcomplexity of the position degree, two kinds of hold group parts are selected asthe measurement objects, such as a square-base part of wire mesh shock-absorber and aflange part with eight holes equally spaced on a circumstance. Firstly, the measuring hardwaresystem was build with the calibration for the CCD camera. Secondly, the grabbing imageswere processed successively by image filtering, edge detection, mathematical morphologicalprocessing, multi-hole segmentation and identification, linear-feature extraction, and centralpositioning, where the neighborhood average is selected for image filtering, the Sobeloperator for edge detection, the regional growing and regional description for multi-hole segmentation and recognition, and the least square method for central location. Finally, theflow chart of software algorithm was determined. The data of image measuring are contrastedwith that of manually measuring and then the error analysis was done. The minor differencebetween the measuring results of the image measurement and that of the manual measurementshows that the image measurement system designed in this dissertation is effective, whichestablishes a good foundation to apply image measurement in the engineering field.