Research On Accurate Measurement Based On Digital Image Processing

With the rapid development of science and technology, control of the product quality in the industrial field demands more accuracy and more automation. Therefore, how to improve the accuracy has become a practical problem. This thesis, based on the system of defect inspection in polymeric film, studied on how to use the technology of digital image processing to improve the degree of measurement accuracy so as to assure the product quality.Signals of defects in polymeric film under scattered light illumination were obtained by online inspecting system with linear array CCD control circuit. The 1-D defect signals were collected and transferred by PCI data acquisition card, and then the 1-D defect signals were analyzed and processed by computer. Thereby, the defects were precisely figured out, and its geometry can be displayed.To improve the detect accuracy, the noise must be reduced firstly. So the source of noise in the defect signal acquired by the CCD device was analyzed, and a new threshold function based on the wavelet soft and hard threshold denoising was brought forward, which avoid the weakness of soft and hard threshold denoising. This modified algorithm not only can reduce the noise effectively, but also can preserve the defect edge well. Since the defect signal which would be processed was only a small part of the whole signal, a dual-threshold image segmentations method was brought forward. It separated the defect from the background successfully When locating the defect edge, according to the characteristics of the detect signal, a new curve fitting method based on Sobel mask in the direction was brought forward. Compared with the polynomial interpolation and straight line fitting, it can effectively improve the accuracy of edge location and the defect size. Lastly, a method of combined programming of VC++ and Matlab based on Matlab engine was used to realize the multi-resolution analysis for the defect. Meanwhile the amount of code was dramatically reduced and the development efficiency was greatly improved.