用于彩色印刷图像的数码相机颜色特性化方法  

Colorimetric Characterization of Digital Camera for Color Printed Images

作　　者：李心怡[1] 黄庆梅[1] 廖宁放[1] 宋莉 胡维贵[1] 陆爱民 

机构地区：[1]北京理工大学光电成像技术与系统教育部重点实验室,北京100081 [2]华夏视科图像技术有限公司,北京100091

出　　处：《光学学报》2013年第B12期243-246,共4页Acta Optica Sinica

基　　金：国家科技支撑计划（2012BAK02B04）、公安部重点研究计划（2010ZDYJBJLG006）

摘　　要：数码相机用于在线检测图像颜色质量的关键问题是数码相机的特性化。针对这一问题，根据在线印品图像颜色检测的要求，采用IT8.7/3色卡作为色样，基于查找表特性化方法的可靠性，提出采用三维查找表与反距离加权插值算法建立数码相机的特性化模型；并针对特定的印品，实际采样进行算法优化。相机预测精度结果最大色差为4.66，平均色差为2.10。在相机预测精度良好的前提下比较了两台相机特性化后的预测值，最大色差为4.40，平均色差为1.34；并提出了采用图像的平均灰度来评价相机一致性的方法。结果表明，此算法使机器视觉一致性有了显著提高，基本满足要求。The key issue of using digital camera to detect image color quality for on-line detection is the digital camera characterization. According to the testing requirements of online color printed image, using IT8.7/3 color card as color samples, based on the reliability of a look up table characterization method, inverse distance weighted interpolation algorithm is proposed; and specific printed images samples are used to improve the accuracy of algorithms. Comparing the predicted and measured values, the maximum color difference is 4.66, the average color difference is 2.10. After that, comparing the predicted values of two cameras, the maximum color difference is 4.40, the average color difference is 1.34. The average gray value of image is used to evaluate the visual consistency of cameras. The results indicate that the algorithm improves the visual consistency so that the camera could meet the basic requirements.

关 键 词：视觉光学 数码相机特性化 三维查找表 反距离加权插值算法 

分 类 号：TP752[自动化与计算机技术—检测技术与自动化装置]