Research On Image Stitching And Processing Method Of X-ray Phase Contrast Imaging System

As a new method of X-ray imaging,X-ray phase contrast imaging technology has become one of the research hotspots in the field of X-ray imaging,because it can achieve high contrast imaging of weakly absorbing substances and has a higher sensitivity than traditional X-ray absorption imaging technology.However,the X-ray phase contrast imaging system with a single sensor structure has a small imaging field of view,which cannot meet the requirements of the large field of view of the actual system.After relevant theoretical analysis and research,it was decided to realize large field of view imaging through multi-sensor array structure,and then used the image stitching technology to process the acquired images.Finally,obtained a complete large field of view X-ray phase contrast image.The image acquisition part of the system is mainly realized by the fiber optic taper coupling image sensor.However,due to the inherent defects in the process of fiber optic taper pulling,the image was distorted,which affects the quality of the final stitched image.By analyzing the existing distortion correction methods and combining the structural characteristics of the system,this paper decided to use the template-based correction method to achieve the distortion correction of the system,and the final correction accuracy was within 1 pixel error,achieved a good correction result.The goal of image stitching is to obtain a complete large field of view image,and the transition between images should be smooth and natural.However,the images acquired by different sensors will inevitably have some degree of brightness differences,and the direct stitching image will have obvious seam-line,which will seriously affect the overall visual effect of the image.In view of this problem,a seam-line elimination method based on improved median filtering combined with total variation smoothing model was proposed.Firstly,the variance-based global brightness correction method was used to eliminate the uneven brightness between images.Then,the pixel gray value in a certain range on both sides of the seam-line was corrected by the improved median filtering method.Finally,the total variation smoothing model was used to smooth the modified image.The experimental results show that the proposed method can effectively eliminate the seam-line while maintaining the edge details of the image,and the overall visual effect of the image is better.