Point-spread Function Of CCD's Optical Systems And The Application In Sub-pixel Edge Localization

In the mechanical industry, the measuring technique is one of industrial development preconditions. With the unceasing improvement of the quality of machine part, the method of the evaluation and measure to the dimension of the part sets a higher request. Along with computer technology and the CCD solid camera technology rapidly development, non-contact measuring technique is gradually becoming the new trend of development. This measuring technique using computer digital image processing technology carries on processing and analysis to image. its key is digital image processing. Point spread function(PSF) is a core concept in the image processing. which used sub-pixel edge localization, image restoration, super-resolution reconstruction, and so on.This paple according to the scalar diffraction theory, summarized the diffraction-limited optical systems point spread function. but the function is a double integral on the infinite interval, it is difficult to achieve its analytical solution. In this paple, useing the definition and the nature of the Bessel function, and approximation to the pupil function, the point-spread function of optical systems was deduced a Bessel's point spread function, this function was an analytic expression, the simple perspicuity, was advantageous for the computation, suited in CCD camera's imaging system.This paple continues to use the view of the majority literature,its thought that image was likely the object and the point spread function of optics imaging systems convolution. Therefore, the point spread function which established in this paple was applied in sub-pixval edge localization. The object of reflection on the verge of different light, with immutable point spread function fitting them as unreasonable, so we should revises to the establish ed point spread function, researched including the revision parameter's Bessell point spread function and the ideal steps and ladders edge model's convolution, obtains the edge the gradation distribution model. According to the least squares method fitting principle, uses the edge gradation distribution model, established new sub-pixel edge localization method - revised Bessel fitting method. Uses Gauss fitting method and this paple research technique separately to the black and white block, the gage block and axis three kind of object image experiment picture processing, the statistical analysis two method results, indicated the fitting residual of this paple fitting method is smaller than Gauss method, it showed that this paple fitting to be accurate, suited in sub-pixel edge localization of the plane components and the axis class components isoline edge. In this paple,gauss fitting and revised Bessel fitting was used to fit many group of different type and different position's edge, discovered their fitted parameter was a constant and equal to 1. It was found that select the window size according to the edge type of object. When used the non-contact measuring technique measure part of the mechanical,the gray value of part and the background's image as poor as black-and-white block such a big beating ,therefore, this article discussed the gray value difference to fit the parameter influence, discovered that they became the linear relationship. With experimental verification, both revised Bessel fitting and Gauss fitting , the resolution not to be lower 4 micron.This paple does the third big aspect worked has made the estimate visible object size to a certain distance, it was helpful laid aside the CCD camera's position in image gathering. According to the eyelet image formation principle, this article deduced the point on shaft surface and its image point coordinate corresponding relationships, it included three kind of situation, the ideal position, the axis of shaft in the horizontal direction has the included angle and the axis of shaft in the normal direction has the included angle,all of those were static shaft. Used these corresponding relationships, obtained the characteristic point of the shaft surface image formation's movement orderliness when shaft online machining process,then deduce motion blur point spread function.This paple is composed of seven chapers.The first chapter summarized the point spread function definition, nature and the existed types, introduced for the optical system point spread function on the background and purpose, finally elaborated the main content of this paple. The second chapter introduced the theory foundation of the point spread function, summarized the diffraction-limited optical system's point spread function. The third chapter used Bessel function's definition and nature further deduced the point spread function of optical systems, and to its normalized treatment. The fourth chapter used deduced optical system's point spread function applied in sub-pixel edge localization.The fifth chapter estimate visible object size to a certain distance, deduced the point on shaft surface and its image point coordinate corresponding relationships, and deduced characteristic point of the shaft surface image formation's movement orderliness when shaft online machining process.The sixth chapter analysied the image point mode of motion when shaft online machining process and deduced motion blur point spread function.The sevEnth chapter is the part of conclusion,summarized the results of this paple achieved,and carried on the forecast to this paple insufficiency as well as the next step work.