Abstract: Aiming at the image gray phenomenon and halo phenomenon based on single-scale Retinex algorithm, gray phenomenon based on bilateral filtering Retinex algorithm and noise amplification phenomenon. An improved two-sided filtering based on wavelet transform and a modified threshold function denoising algorithm are proposed. The image is decomposed by wavelet transform to obtain the low-frequency and high-frequency coefficients of the image. The low-frequency coefficients of the image are processed by improved bilateral filtering Retinex algorithm, and the high-frequency coefficients are processed by improved threshold function. The enhanced reconstructed image is obtained by inverse discrete wavelet transform. The reconstructed image is reconstructed like segmented linear transformation, enhance image contrast. The experimental results show that this method avoids the phenomenon of image gray and halo, and effectively removes noise, rich details and strong contrast, which lays the foundation for image follow-up processing.

Key words: wavelet decomposition, improved bilateral filtering, Retinex algorithm, improved threshold function, discrete wavelet inverse transform, piecewise linear transform

摘要： 针对基于单尺度Retinex算法产生的图像泛灰现象和光晕现象、基于双边滤波Retinex算法的泛灰现象及噪声放大现象。提出基于小波变换的改进双边滤波的Retinex图像增强算法和改进阈值函数去噪算法。该方法对图像进行小波分解，获得图像的低频和高频系数；采用改进双边滤波的Retinex算法对图像低频系数进行处理，采用改进阈值函数对高频系数进行处理；采用离散小波反变换得到增强后的重构图像；对重构图像进行分段性线性变换，增强图像对比度。实验结果表明，该方法避免了图像泛灰和光晕现象，并有效去除了噪声，细节丰富，对比度强，为图像后续处理奠定基础。

关键词: 小波分解, 改进双边滤波, Retinex算法, 改进阈值函数, 离散小波反变换, 分段性线性变换