The Research And Improvement Of Some Medical Image Enhancement Algorithms

Medical imaging is the important basis of modern medical diagnostics. However, due to thelimitations of the hardware systems, medical images obtained directly from the medicalinstruments may present poor characteristics such as poor resolution or low contrast.Therefore, in order to improve the accuracy of medical diagnosis, and let the output image tomeet the specific application, it is greatly significant to propose new medical imageenhancement technologies.In this paper, firstly, the domestic and international status as well as the basic methods ofmedical image enhancement is described. And after analyzing the problems and shortcomingsof the existing medical image enhancement algorithms encountered in practical applications,two novel medical image enhancement methods are proposed. The main content andinnovation of this article are as following:I. A novel algorithm named Median-Separate Histogram Equalization is proposed. Thisnew algorithm maintains the advantages of contrast enhancement from the traditionalhistogram equalization. It is proved by mathematics that the new algorithm has betteradaptability and performance than the others. It can maintain the brightness of the originalimage while doing contrast enhancement. The simulation results show that the algorithm isbetter than the original one;II. In the foundation of Human Visual System decomposing image as well as thecombination of improved histogram equalization method and alpha weighted quadratic filter,a new algorithm that maintain the background details of image while doing contrastenhancement is proposed. This new algorithm preserves the advantages of histogramequalization and alpha weighted quadratic filtering, so it can prevent the background area ofimage from being destruction while enhancing the image contrast. The simulation resultsshow that this method has better contrast enhancement effects，and has the ability to maintainthe details of the background areas at the same time.