Research On High Accuracy Dimensional Measurement Method Of Industrial CT Image

Industry CT image measurement is a measure method, which the study object is CT image and measures the inner geometric size according to the dealing with industry CT image and the measurement of image's geometric size. Up to now it has been widely used in the field of industry test and medical diagnosis. In recent years, measuring by industry CT is one of the important study directions from qualitative measurement to quantitative measurement. With the high requirement to the measuring accuracy by practical application, the present measure way couldn't satisfy the practical need, so the measure method with high accuracy is much of value in application. And the main study contents of the paper are as follows:Firstly it propose a improved method based on the Zernike moment which has the ability of thinning edge and consider the first derivative model of the edge along the gradient direction of the edge. That boosting the measuring accuracy of image's geometric size is mainly determined by the accuracy of CT industry edge localization. the accuracy of the traditional pixel-level edge detection way is not high, so it presents sub-pixel edge detection technology, which is focus on the study of sub-pixel edge detection method based on Zernike moment and CV model, and point that Zernike moment have two shortcomings: not single edge and not consider the first derivative model of the edge along the gradient direction of the edge, then propose a limited non-optimal suppression and the intensity of edge improved method based on the Zernike moments.Secondly it comparatively analyzes the advantages and disadvantages of the improved Zernike moment and CV model edge detection method from the theory aspect and experiment aspect. From the aspects of the speed, measurement accuracy, anti-noise ability, edge continuity, the influence from the edge image topology of the testing object and the adaptability of edge detection with multi-targets, it theoretically analyzes and experimentally comparisons of Zernike square and the CV model, which aims to get the edge detection method that more adapt to industry CT image.Thirdly it proposes an improved method based on the Zernike moment which has the ability of correction edge location error at the neighborhood of the junction. The method is based on the difference of the angle factorφand linear fitting method.Fourthly it points that Zernike moment have a shortcoming: limitation in detection the edges of curves, then propose an improved method based on the Zernike moments through penalty function of measured value. The ideal model of the sub-pixel edge detection is based on ideal edge line, which is not satisfied of the edges of curves. The improved method propose penalty function of measured value when the detection target is circles. The radius measuring method is firstly edge detection based on Zernike moment method, and edge thinning, then get circle radius through least squares fitting, after that calculation error compensation factor, get revised circle radius.Fifthly it measures length and thickness based on the Zernike moments and get the adapt range according to the experiments. The way of length measuring is firstly to detect edge based on Zernike moment method, and then get the regions of interest through mouse, and get linear equation through least squares fitting, after that get the length through calculating the distances between the parallel. The way of thickness measuring is firstly to detect edge based on Zernike moment method, and get the regions of interest through mouse, then use interpolation, after that get the thickness through calculating the minimum distance.The paper has a detailed study on Zernike moment and CV model sub-pixel edge detection method. Based on the analysis of the disadvantage of the present way, it proposes the improving way and establishes the size measuring method with high accuracy and testifying the measuring accuracy and adapts range according to the experiments. Different from the present way, the measuring method in the paper is based on the sub-pixel edge, which break through the limit from CT image resolution ratio to the measurement accuracy, and make it possible that CT image with low resolution ratio realize the high accuracy measurement, so it has a meaningful value in practical engineering application.