Geometry Correction For X-ray Image

Using X-ray for industrial inspection and medical pathology analysis has been widely used in our country. People apply machine vision technology to the X-ray real-time imaging system for computer aided analysis and even completely replacement of the human analysis. One of the important things in analyzing X-ray images is locating the object position accurately, but images from X-ray apparatus are often geometric distorted ,which leads to the analysis errors. So it is necessary to correct such images.In this paper I consider the causes of the geometric distortion in X-ray images, and apply the camera calibration into the X-ray image correction. It will guarantee a precise quantitative analysis for X-ray detection. The thesis is mainly as follows:Firstly, this paper considers how the X-ray apparatus produces an image. Most of the problem in this process only blurs the image, except the camera at the end of the apparatus which takes photo and transport it to the computer. The defect in the camera is the main reason to the X-ray image distortion. Images from the cheaper cameras are distorted obviously, especially in the wide-angle case. So I use the camera calibration method for X-ray image geometric correction.Secondly, I design a new calibration board according to the X-ray environment. The calibration board in this paper is made from small metal pieces pasted on an organic glass board. The X-ray apparatus will produce a image with high contrast with this board.Finally, I use the sub-pixel Harris corner detection method while extracting the calibration points. The camera model is pinhole model. The distortion model includes radial distortion and tangential distortion. The calibration method is based on the paper from Zhang Z.Y. Having obtained the parameters of the camera, the X-ray image can be corrected by searching the undistorted pixel position in the distorted image.