改进小波亮度融合的低照度图像增强算法

Low illumination image enhancement algorithm based on wavelet transform

针对普通小波分解算法应用到夜间低照度图像增强时出现无法提取出边缘光滑特征点、且亮度拉伸曝光等问题,提出了一种改进小波亮度融合的低照度图像增强算法.在小波变换对夜间低照度图像进行频域变换的过程中分别提取出图像的低频和高频分量,并对高低频分量分别处理.对小波分解后形成的低频成分使用加入权值的引导滤波,得到边缘增强的低频分量.将高频成分基于不同的区域范围进行特性融合,得到细节均匀统一的高频分量.将处理后的分量进行亮度提取与融合,最后转入RGB空间得到增强图像.仿真实验结果表明,该算法在均值、信息熵、峰值信噪上相较于直方图均衡算法提高了21.715%、13.531%、73.768%,相较于小波变换提高了85.456%、26.014%、19.621%.

A new low-illumination image enhancement algorithm based on improved wavelet brightness fusion was presented,which can not extract smooth edge features and stretch exposure when ordinary wavelet decomposition algorithm was applied to night low-illumination image enhancement.The low-frequency and high-frequency components of the night low-intensity image were extracted and processed separately during the frequency domain transformation of the night low-intensity image by the wavelet transformation.The low-frequency components formed by the wavelet decomposition were filtered by a guided filter with weights to get the edge-enhanced low-frequency components.The high frequency components were fused based on different region ranges to obtain uniform high frequency components with uniform details.The processed components were extracted and fused into RGB space to get the enhanced image.The simulation results showed that the algorithm improved 21.715%,13.531%,73.768%in mean,information entropy and peak signal-noise compared with the histogram equalization algorithm,and 85.456%,26.014%,19.621%compared with the wavelet transform.