Study On Edge Detection And Its Application In Concentricity Measurement

With the development of image processing technology and large area, high resolutionimage sensor has been launched, image measurement system is widely used in precisionmeasuring area, and gradually become an important testing method in this field. Edgedetection is the base of image measurement, the accuracy position of boundary directly affectsmeasuring accuracy. In this paper, an in-depth study on edge detection has been done, and itcan be used in the concentricity measurement of microminiature parts. In this paper, the mainwork can be divided into the following parts:First of all, discussed the importance and significance of edge detection, the developmentsituation of edge detection in domestic and abroad were introduced in detail and simplyclassified edge detection algorithm, analyzed the difficulties in edge detection and its futuredevelopment trend.Secondly, studied the related content of digital image processing, include image graylevel, several commonly used image smoothing algorithm, binarization algorithm, pixel edgedetection algorithm and compared the performance and their use situation, because the edgedetected by the classical edge detection algorithm lack of precision, unable to meet therequirements of high precision measurement, several sub-pixel edge detection algorithm arestudied which are commonly used, including gray moment, Zernike orthogonal moment, andspatial moment, At last using C++program to implement them, compared the performance ofthese algorithms in accuracy according to the experimental results.Thirdly, researched the circle detection algorithm, elaborated the Hough transformclearly, as the Hough detection based on gradient information can not locate the circlecorrectly, so used the least squares fitting to correct it, Hough transform has disadvantages ofinvalid accumulation and low speed, in this paper gave a morphology combined with edgetracking, multiple least-squares fitting method to detect circle in this paper.Fourthly,it established the image measurement system in which Bi-telecentric lens is thecore equipment, and analyzed light source, industrial camera, image sensor, effectmeasurement system, and made the choices. Finally, designed a simple and fast calibration method on the basis of the characteristics of double telecentric lens.The last, it builds a measurement system in the laboratory existing image measurementenvironment. Aimed at the target being measured, using image measurement system tomeasure and contrast to its actual value is to prove the feasibility and effectivity of usingimage measurement technology for the concentricity measurement, to prove a certainpractical values and the advantages of simple, convenient and high precision of this method.With the development of the technology of image measurement and the improvement to themethods of measurement is expected to be used in real-time, on-line, non-contact, fastdetection in actual production process.