一种基于小波相位滤波及视觉非线性的医学图像自适应增强新方法

A Novel Mehod for Image Enhancement of Medical Images Based on Wavelet Phase Filter and Nonlinear Human Visual Properties

针对经典方法对 Ｓ Ｎ Ｒ 低的医学图像存在噪声过度放大及伪像产生问题，本文在精细尺度上，根据信号与噪声的 Ｗ Ｔ 相位在相继尺度上关联性的不同进行去噪．在大尺度上，则采用 Ｓｅｍｉｓｏｆｔ 阈值法对 Ｄ Ｗ Ｔ 系数进行快速缩减去噪．根据人眼的视觉特性对 Ｗ Ｔ 系数的增益进行非线性的自适应控制．较之经典方法，本文方法具有增强图像视觉效果佳，无伪像产生的优点．且在噪声抑制、保边沿及增强各种细节上效果良好．

The goal of this research was to present an approach which address both de noising and contrast enhancement.1)2D wavelet transform (WT).2)At finer scales,wavelet phase filter was used in WT domain to reduce noise and at coarse scales,and semisoft wavelet shrinkage technique was used to furtrher reduce noise.3)Based on human visual properties,the coefficients of 2D WT of image were emphasized by a generalized adaptive gain nonlinear operator.4)Image reconstruction.Preliminary results show that the proposed approach can make unseen or barely seen features of an image more obvious without generating artifacts and enhance the contrast of region of interest in images while reducing the enhancement of noise.