Research on image preprocessing based on 3D reconstruction of sequential X-ray images

Abstract

The original X-ray images have problems such as high noise, narrow gray range, low contrast, and blurred edges, which seriously affect the reconstruction accuracy when using them for 3D reconstruction. Therefore, this paper explores the image preprocessing method of three-dimensional reconstruction of sequential X-ray images, an improved median filtering algorithm with adaptive neighborhood threshold is proposed, which effectively solves the problem of rapidly processing high pulse noise in X-ray images while retaining the edge detail features of the target, and then highlight the difference between the image target and the background through the histogram equalization method. Finally, the Canny operator is used to extract the edge of the target. Experiments show that the preprocessing process can effectively remove the noise of the X-ray image, better highlight the edge details of the target, and can extract clear edge information, which is convenient for subsequent 3D reconstruction work.