基于稀疏查找表的色彩变换插值算法研究

ON COLOUR TRANSFORMING INTERPOLATION BASED ON SPARSE LOOKUP TABLES

使用均匀查找表执行色彩变换的传统方法会占用大量内存资源。针对均匀查找表大小随输入维数增长而成指数增长的特点,提出稀疏查找表方法,通过将整个设备空间分层为不同维数的子空间,在包含重要色彩组合的区域完成合理密度的抽样,优化查找表大小,更有效地节约内存空间。研究表明基于稀疏查找表的插值算法比传统方法更适用于三通道以上的多通道设备色彩变换,是新一代色彩管理系统的一个重要改进。

The conventional approach that using uniform lookup table （LUT） for implementing colour conversion demands must consume vast memory space. By analyzing the characteristic that the LUT size increases exponentially with the input dimension,the approach based on sparse lookup tables are developed. By stratifying the whole device space into subspaces of various dimensions, this algorithm is able to achieve reasonably dense sampling in regions of the device space that contain important combination of colorants, optimize the LUT size so that to save memory space more efficiently. The research conclusion reveals that the interpolation based on sparse LUT is more adaptive than conventional approach in the eolour transformation of device with more than three channels. This new approach is an important improvement of the next generation eolour management system.