A Kernel Based Algorithm for Fast Color-to-Gray Processing
A Kernel Based Algorithm for Fast Color-to-Gray Processing
复制标题
一种基于内核的快速彩色到灰度处理算法
DOI：
10.1109/cisp.2008.411
复制
发表时间：
2008
期刊：
影响因子：
--
通讯作者：
Zhuo Sun
中科院分区：
文献类型：
--
作者：
Yi Zhang;Chun Chen;Jiajun Bu;Mingli Song;Chenkang Pang;Zhuo Sun
彩色转灰度是图像处理中的经典操作。本文提出了一种基于核的方法来执行彩色到灰度的操作。值得注意的是，在有意义的彩色图像中，仅仅每个像素的三个颜色分量的线性组合不足以描述每个像素内或像素之间的亮度和色度之间的蕴含关系。这种方法的动机来自于内核主成分分析。引入核方法来间接捕获和隐式提取非线性关系。与传统的内核PCA不同，这种方法可以更快地执行颜色转换，同时尽可能保留彩色图像的显着特征。设计最佳核函数是为了增强我们方法的有效性。实验结果表明，我们的方法在质量上具有明显的竞争力，并且在性能上具有很大的优势。
Color-to-gray is a classical manipulation in image processing. In this paper, a kernel based method is presented to carry out color-to-gray operation. It is noticeable that, in a meaningful color image only the linear combination of the three color components of every pixel is hardly enough to describe the implication relation between the luminance and chrominance in every pixel, or between pixels. The motivation of this approach comes from the kernel principle component analysis. The kernel method is introduced to capture and extract implicitly the nonlinear relationships indirectly. Different from the conventional Kernel PCA, this approach can perform the color conversion much faster while preserving salient features of the color image as much as possible. An optimal kernel function is designed to enhance our method's effectiveness. Experiment results shows that our method is appreciable and competitive in quality with great advantage in performance.