融合亮度信息与色度信息的图像梯度提取方法

Novel Gradient Extraction Based on Luminance Information and Chrominance Information

梯度信息对人眼视觉感知具有重要作用,准确的图像梯度描述对于图像边缘提取、图像增强、图像分割等都有着极为重要的意义。传统的梯度提取方法通常仅利用图像的亮度信息,当图像颜色发生改变而亮度变化不大时,传统方法获得的梯度数值则会与人眼的视觉感知产生矛盾。提出一种融合亮度信息与色度信息的图像梯度提取方法。该方法在保留亮度梯度信息的基础上,引入基于人眼视觉模型的CIE-L*a*b*颜色模型获得色度梯度信息,而后将归一化的亮度梯度与色度梯度进行了梯度融合。实验结果表明,该方法可充分利用亮度信息与色度信息,产生更符合人眼视觉感知的梯度描述。

Gradient plays an important role in human visual perception. Accurate gradient descriptor is critical for many image processing applications such as edge detection,image enhancement and image segmentation. It is common that only luminance information is utilized by traditional gradient descriptors. When two different colors have similar luminance in the image,the gradient values calculated using traditional gradient descriptor are not consistent with human perception. In this paper,a novel gradient extraction approach is proposed,which depends on both luminance information and chrominance information. Besides the existing luminance gradient descriptor,a chrominance gradient is calculated using the CIE-L*a*b*color model,which is established based on the human visual model on color difference. Then the luminance gradient and chrominance gradient are normalized and fused together. As shown by large amount of experimental results,the proposed gradient descriptor can fully utilize luminance information and chrominance information and achieve a good gradient description consistent with human perception.