基于多项式回归模型的液晶显示器特性化

为了实现LCD液晶显示器精确的色度特性化,用多项式回归算法建立了液晶显示器RGB空间到CIEXYZ空间的转换模型,详细研究了多项式回归模型不同训练样本、不同项数以及阶数对液晶显示器特性化的影响。实验结果表明,项数小于9时,模型预测的精度较低,项数大于9时模型精度逐渐提高,用三阶三元十四项多项式可以得到平均误差为0.6623左右的高精度预测模型。

Colorimetric Characterization for LCD Based on Polynomial Regression Model

In order to realize accurate colorimetrie characterization for a LCD, several RGB to CIEXYZ color conversion models using polynomial regression algorithm were established. The influence of term number, order number of polynomial regression model and specimen on characterization precision of display were analyzed with experiment. The results showed that the model accuracy was lower when the items of polynomial were less than 9 and the accuracy was improved when the items were 9 or more. The polynomial regression prediction model with tree orders, three elements, and fourteen terms has higher accuracy, with difference about 0. 6623.