Research On Medical Image Enhancement Algorithms

With the appearance and development of medical imaging technology,digital medical imaging has become the main medium of medical information analysis and transmission.Nowadays,medical image processing and analysis play an important role in assisting physicians.However,many unfavorable factors affect medical diagnosis,such as electrical noise,photon noise and X-ray scattering in the imaging process.Medical images are ubiquitous in the areas of low contrast,narrow dynamic range and asymmetric intensity distribution.Therefore,the application of image enhancement technology in the medical field has an indispensable significance.This paper studies several widely used image enhancement algorithms,and applies them to medical images to summarize the enhancement effects and emerging problems.The main contents are as follows:(1)Several extensively used image enhancement algorithms are applied to medical image enhancement to analyze the effect.(2)Three kinds of variational Retinex algorithm are studied,and each method introduces different regularization prior constraints for the illumination component and the reflection component.(3)Aiming at the above problems,a Retinex algorithm(CWGFR)based on the Canny operator weighted-guided filtering is proposed.Firstly,Canny operator is used to accurately estimate the edge weight of the weighted guided filter.Secondly,the illumination is estimated by the weighted guided filter.Finally,the reflection is calculated to obtain the output image.The experimental results show that the proposed algorithm can enhance the image contrast.(4)To handle above problems,a new variational Retinex algorithm(SSVR)based on space adaptive is proposed.The algorithm preserves the edge information by using a surround suppression mechanism and enhances the regularization constraint to smooth the uniform area.Finally,the usage of the split Bregman iterative method is to accelerate the solution.The algorithm in this paper and a variety of enhanced algorithms are achieved on different data sets for comparative experiments.