A Medical Image Enhancement Method Based on Improved Multi-Scale Retinex Algorithm

摘要

Enhance the dark area information and image contrast in medical images, and compress the dynamic range of medical images. Methods: A medical image enhancement method based on multi-scale Retinex method was proposed and improved. Under three standard deviations, the Gauss environment function was obtained. Then three different error functions were used to convolute the medical image, and the results obtained under three standard deviation scales were weighted averaged. Finally, the output gray value was modified to obtain the results that can be used for display. Results: The improved MSR algorithm can not only enhance the medical image with low contrast, but also compress the dynamic range of the image. It can significantly improve the information of the medical image in dark area. The improved MSR algorithm can significantly improve the information content of dark area image, and the enhanced information entropy is more than twice that of the original image. This method overcomes the shortcomings of conventional medical enhancement methods and can meet the requirements of doctor' clinical diagnosis. The experimental results show that compared with the traditional Retinex image enhancement algorithm, this method effectively improves the visual effect of medical images and enhances the details and color information of the images.