Research On Profile Connection Measurement Based On Image Processing

Mechanical parts precision detection is an important part of the manufacturing, and now, the demand of people for the part of detection accuracy, automation and efficiency have become increasingly with the improvement of machining accuracy and complexity, which has a higher requirement for the detection accuracy. When people need to do the real-time measurements on the complex shape parts, the traditional detection method appears powerless. As the logarithmic spiral described in this dissertation, can not detect its accuracy with the traditional detection methods, so the author make use of the image processing methods to detect, and find that the method is feasible through the experiments.This dissertation introduces the detection of parts of the image processing steps and methods: introduces the image acquisition device first; doing the image pre-processing,such as image type conversion,noise removal and binarization after acquiring the image; boundary extraction; fitting the parts curve for accuracy analysis at the last.When doing the edge extraction on the differentiation of several commonly used operators are analyzed and compared with the effects of mathematical morphology, finally adopts the method of mathematical morphological to carry out the edge extraction . Based on the characteristics of the image a dedicated system is designed for the precision analysis, using the boundary-tracking detect the line, and then strikes the center of parts according to the intersection of three straight line. Finally adopts the method of curve fitting to do the logarithmic spiral fitting, and then do a comparative analysis with the standard components after fitting the curve.In this dissertation, we found image processing methods with surface accuracy of detection of profile connections is possible through the experiment, which achieves the purpose of time-saving effort and convenient. It is an important future direction for the development of detection methods which opens a way for the non-contact detection.