光谱学与光谱分析, 2020, 40 (2): 454, 网络出版: 2020-05-12   

用聚类分析的方法分类观察者颜色匹配函数

Classification of Color Matching Functions with the Method of Cluster Analysis
作者单位
北京印刷学院印刷与包装工程学院, 北京 102600
摘要
色觉正常的观察者, 尤其是不同年龄段的观察者, 颜色分辨能力具有较大的差异。 为了分类色觉正常观察者的锥细胞光谱响应, 补充现有CIE1931, CIE1964和CIE2006颜色匹配函数, 基于Stiles&Burch数据集的47个观察者函数(10°视场)和CIE2006代入20~80岁(10°视场)观察者不同年龄产生的61个函数(共108个颜色匹配函数), 用聚类分析方法中的k-medoids算法, 计算距离选用欧氏距离的平方, 将108个颜色匹配函数在(λ), (λ), (λ)三个通道分别聚为5类, 共产生了5×5×5=125个类别。 将108个颜色匹配函数作为108个“个体观察者”, 在iPad显示设备上呈现CIE推荐的17个颜色, 用108个颜色匹配函数和125个类别颜色匹配函数比较, 计算17个颜色的CIEDE2000色差平均值, 以最小色差值作为评价标准, 从125个类别中挑选出10个分类观察者颜色匹配函数表征观察者锥细胞光谱响应。 108名“观察者”中有77.8%的观察者属于这10个分类。 在iPad显示设备上显示CIE推荐的5个颜色(灰、 红、 黄、 绿、 蓝色), 组织30名(20~25岁)年轻观察者和17名(61~74岁)老年观察者, 在Quato专业显示器上依次匹配5个目标色, 共采集到158组, 即790个(=158组×5个色中心)颜色匹配实验数据。 为将这158组数据(视为158名“观察者”)进行分类, 分别将10个分类颜色匹配函数代入计算, 以计算得到5个颜色CIEDE2000色差平均值最小的颜色匹配函数, 作为158名观察者相对应的分类, 在10个分类观察者函数中优选出8个颜色匹配函数BIGC-1, BIGC-2, …, BIGC-8。 进一步地, 用优化建立的分类颜色匹配函数检验课题组前期基于近同色异谱色样对开展的目视色差比较实验, 结果发现BIGC-3适用于年轻观察者, BIGC-5适用于老年观察者, 同时计算得到的STRESS值也较其他颜色匹配函数的结果有所降低。
Abstract
Larger color discrimination difference exits among observers with normal color vision, especially those of different ages. In order to classify the cone fundamentals among normal color observers, the 108 color matching functions (CMFs) including 47 Stiles&Burch CMFs and 61 CMFs computed by CIE2006 model were respectively classified into 5 categories in (λ), (λ), (λ) three channels by using the k-medoids algorithm of clustering analysis method as well as the square of Euclidean distance and a total of 5×5×5=125 categories were generated. Taking the 108 CMFs as 108 “individual observers”, the 17 color centers recommended by CIE were displayed on the center of iPad and 108 CMFs were compared with 125 categorical functions, then the average of 17 colors’ CIEDE2000 color differences were calculated. Finally 10 categorical CMFs were selected from those 125 categories to represent the spectral response of human cone cells. The results indicated that 77.8% from the 108 “real observers” were satisfied, which regarded the obtained minimum color difference as the objective function. As the target colors, CIE recommended 5 colors (gray, red, yellow, green, blue) were presented on the iPad and 30 young observers aged 20 to 25 and 17 old observers aged 61 to 74 were organized to match 5 color centers correspondingly on Quato display. Therefore, 158 groups, 790 color data (each group includes 5 color centers) were obtained and then categorized by computing CIEDE2000 color difference using 10 categorical CMFs. The CMFs possessing the minimum color difference value were assigned as the corresponding classification of 158 observers and finally 8 out of 10 categories were selected and named BIGC-1, BIGC-2, …, BIGC-8, which were used to test the results of paired comparison experiment based on metameric color samples in our previous study. The obtained results show that BIGC-3 CMFs worked well for young observers while BIGC-5 CMFs were suitable to old observers. Additionally, the calculated results of STRESS value were also lower than the results computed by other CMFs.

黄敏, 郭春丽, 何瑞丽, 习永惠. 用聚类分析的方法分类观察者颜色匹配函数[J]. 光谱学与光谱分析, 2020, 40(2): 454. HUANG Min, GUO Chun-li, HE Rui-li, XI Yong-hui. Classification of Color Matching Functions with the Method of Cluster Analysis[J]. Spectroscopy and Spectral Analysis, 2020, 40(2): 454.

本文已被 1 篇论文引用
被引统计数据来源于中国光学期刊网
引用该论文: TXT   |   EndNote

相关论文

加载中...

关于本站 Cookie 的使用提示

中国光学期刊网使用基于 cookie 的技术来更好地为您提供各项服务,点击此处了解我们的隐私策略。 如您需继续使用本网站,请您授权我们使用本地 cookie 来保存部分信息。
全站搜索
您最值得信赖的光电行业旗舰网络服务平台!