An Improved Algorithm Based on Retinex Theory for X-Ray Medical Image

Abstract

Retinex theory combined the elements of images and visual.This paper improved the Retinex-based medical image enhancement method, It can get better brightness by using the neural network logarithmic The S-shaped LogSig transfer function instead of the original MSR logarithm function. Based on this, the paper presents a composite LRA (LogSig Retinex Algorithm) algorithm, and analysed the shortcomings of the original Retinex algorithm applied to the X-ray medical image analysis, described the advantage of the composite LRA algorithm is better than traditional Retinex algorithm on the X-ray medical image. Experimental results show that the improved Retinex algorithm can achieve not only low-contrast medical image enhancement, but also the dynamic range compression of the image, can significantly improve the information of the medical image of the dark area. It has practical significance for clinical diagnosis.