结合人眼亮度调制传递函数的形态学锐化算法

Morphology Sharpening Algorithm Combining Modulated Transfer Function of Human Eye Luminance

针对常规彩色图像锐化方法存在的局限性,提出一种空间域锐化算法,其中包括数学形态学运算,利用高斯滤波器分离图像的低频成分和高频成分,强化图像高频成分,调整图像低频成分,将两者合成为新的输出图像。由于形态学开运算和闭运算分别具有自适应极大值滤波和极小值滤波的特点,因此将开闭组合运算用于图像高频成分,自适应地突出图像局部细节,再根据人眼亮度调制传递函数模型,对图像的低频成分进行全局调整,从而进一步改善图像的整体视觉效果。结果证明,该方法在RGB彩色空间和色调/亮度/饱和度感知彩色空间内的锐化效果都好于常规锐化方法。

Aiming at limitation of classical methods of color image sharpening,a kind of novel spatial sharpening algorithm is proposed,which includes mathematical morphology.Gaussian filter is used to separate low frequency and high frequency components of image.High frequency component is intensified,low frequency component is modulated,and the two components are composed into new output image.Morphology open and close operations are composed to apply to high frequency component of image because the two operations have separately adaptive maximum and minimum filtering feature,consequently local detail of image is automatically highlighted.Then modulated transfer function model of human eye is adopted to low frequency of image,and whole visual appearance of image is further improved.Experimental results show that the method is better than classical image sharpening algorithms in RGB color space and hue/illuminant/saturation perceptional color space.