摘要
利用四面体插值算法进行色彩校正时,通过样本集直接建立起来的三维查找表往往是不均匀的,为了生成符合ICC规范的Profile,必须通过插值得到均匀的Profile数据。在均匀化过程中,耗时最严重的是如何准确定位插值颜色点所在的四面体。从数据的邻域相关性,除了基于历史的局部搜索法,同时从粗定位和细定位角度提出了基于辅助表的快速定位算法。实验结果证明任何一种加速算法都有效缩短了Profile数据生成的时间,同时结合两种加速算法的效果更佳。
In the process of color calibration using tetrahedral interpolation, 3D-LUT created directly by color patch sample is usually not uniform, which contradicts the ICC profile specification and makes it necessary to obtain the uniform profile data by interpolation. However, the time-consuming location of proper tetra for interpolation for uniform data is a serious problem for creating the profile data. Two fast searching algorithms, namely, local searching based on history and fast location using auxiliary table, are presented from the perspective of neighborhood correlation and location at the rough and detailed level. Experimental results indicate either of the fast searching algorithms can effectively shorten the time of creating profile data and the combination of two fast searching algorithms can perform better than either of them.
出处
《计算机科学》
CSCD
北大核心
2007年第2期258-259,276,共3页
Computer Science
基金
国防预研基金资助项目(413160501)
西安电子科技大学研究生创新基金资助课题(05008)
