在均匀颜色空间中实现彩色图像的颜色量化  

Color quantization of color images in a homogeneous color space

作　　者：任智斌[1] 隋永新[1] 杨英慧 杨怀江[1] 

机构地区：[1]中国科学院长春光学精密机械与物理研究所应用光学国家重点实验室,吉林长春130022 [2]辽宁营口中专,辽宁营口115000

出　　处：《光学精密工程》2002年第4期340-345,共6页Optics and Precision Engineering

基　　金：国家自然科学基金(No.60178018);吉林省杰出青年基金(No.20005502)

摘　　要：颜色量化是将原有图像中的多种颜色根据人的视觉效果归类为较少的颜色 ,从而用这些较少种类的颜色重新生成一幅新的图像 ,使量化后的图像与原图像的差别即量化误差最小。采用了将RGB非均匀颜色空间变换到L a b 为基准的均匀颜色空间的方法 ,在L a b 均匀颜色空间中对彩色图像进行颜色量化 ,从而解决了量化误差相对人眼不均匀的问题。首先将彩色图像的像素数据保存在单链表中 ,然后对链表进行扫描 ,并把本次扫描得到的色差最小的两个节点合并成一个新的节点。经过足够的动态运算后可得到量化处理后的图像调色板。实验结果表明 ,本方法具有普遍性、唯一性 ,可以减小量化误差 。Color quantization is to classify the colors of an image to a few classes of colors according to the visual effect of human eyes. So these few colors can be used to creat a new image and keep the color quantization errors between the new image and the onginal one to be the least.This paper gives the quantization of a color image in a horrogeneous color space of L *a *b *, which is transformed from a nonhomageneous color spece of RGB. Therefore the problem of nonhomogeneous chromatisom to human eyes can be solved. Firstly,, the data of a color image were saved into a single chain list. Then the list was scanned, finding two nodes whose chromatism is the least. Finally,the two nodes were combined into a new node. After sufficient dynamic calculation we got the palette of the color image quantized. The experimental results show that minishing the color quantization errors and improving its veracity are the catholicity and uniqueness of this method.Key words:

关 键 词：颜色空间 颜色量化 彩色图像 计算机 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术] O432.3[自动化与计算机技术—计算机科学与技术]