基于AKAZE-FLANN算法的色卡自动提取方法研究

Research on Automatic Color Checker Extraction Based on AKAZE-FLANN Algorithm

为学习未知光照条件下失真彩色图像和标准光照条件下目标视觉感知的颜色外观之间的映射模型,针对现有方法大多采用手动或半自动方式提取色卡中颜色信息效率较低的缺点,结合AKAZE特征检测、快速最近邻搜索算法(FLANN)与随机抽样一致匹配算法(RANSAC),提出一种色卡识别与颜色信息自动提取的方法。为测试方法的有效性,具体以115幅采集图像作为测试集,将该算法与人工标定、基于SIFT和SURF特征检测算法进行对比实验,结果表明该方法的颜色信息提取准确率达97.39%,具有一定的准确性和鲁棒性。

Aiming at the disadvantage of low efficiency of the existing method by which color information is extracted from color checker by manual or semi-automatic methods,this paper proposes a method of color checker recognition and automatic color information extraction which integrates AKAZE feature detection,Fast Library for Approximate Nearest Neighbors(FLANN)algorithm and Random Sample Consensus(RANSAC)algorithm.The proposal of this method is also to learn the mapping model between distorted color images under unknown illumination conditions and the color appearance of target visual perception under standard illumination conditions.In order to test the effectiveness of the method,115 acquired images are taken as a test set,and the proposed algorithm is compared with manual calibration and detection algorithms with SIFT(Scale-Invariant Feature Transform)and SURF(Speeded Up Robust Feature)features.The results show that the accuracy rate of color information extraction by the proposed method reaches 97.39%,with a certain degree of accuracy and robustness.