仪器仪表学报
儀器儀錶學報
의기의표학보
CHINESE JOURNAL OF SCIENTIFIC INSTRUMENT
2014年
12期
2874-2880
,共7页
数码相机%颜色空间转换%颜色质量检测%多项式模型%分空间补偿模型
數碼相機%顏色空間轉換%顏色質量檢測%多項式模型%分空間補償模型
수마상궤%안색공간전환%안색질량검측%다항식모형%분공간보상모형
digital camera%color space conversion%color quality detection%polynomial regression%sub-space compensation model
基于数码相机的印刷品颜色质量检测首先面对的问题是颜色空间的设备相关性,解决这一问题的方法是进行数码相机的特征化,即将设备相关的RGB颜色值转换成设备无关的CIELab色度值。根据数码相机颜色空间的特性,提出基于多项式回归的子空间自适应补偿模型算法。该模型首先对RGB颜色空间进行划分,在每个子空间中建立多项式回归模型获得预测色度值及其转换精度,然后根据该转换精度与精度阈值间的关系确定是否需要对预测色度值作优化补偿处理。对于转换精度低于阈值的子空间,将其预测色度值与实测色度值的差值拟合为RGB颜色值的多项式函数,获得预测差值的多项式模型,并利用色度差值对其色度值进行补偿。实验表明,该方法能够有效补偿不同子空间颜色特性对转换精度的影响,大大提高了颜色空间转换的精度,满足颜色质量检测对颜色空间转换精度的要求。
基于數碼相機的印刷品顏色質量檢測首先麵對的問題是顏色空間的設備相關性,解決這一問題的方法是進行數碼相機的特徵化,即將設備相關的RGB顏色值轉換成設備無關的CIELab色度值。根據數碼相機顏色空間的特性,提齣基于多項式迴歸的子空間自適應補償模型算法。該模型首先對RGB顏色空間進行劃分,在每箇子空間中建立多項式迴歸模型穫得預測色度值及其轉換精度,然後根據該轉換精度與精度閾值間的關繫確定是否需要對預測色度值作優化補償處理。對于轉換精度低于閾值的子空間,將其預測色度值與實測色度值的差值擬閤為RGB顏色值的多項式函數,穫得預測差值的多項式模型,併利用色度差值對其色度值進行補償。實驗錶明,該方法能夠有效補償不同子空間顏色特性對轉換精度的影響,大大提高瞭顏色空間轉換的精度,滿足顏色質量檢測對顏色空間轉換精度的要求。
기우수마상궤적인쇄품안색질량검측수선면대적문제시안색공간적설비상관성,해결저일문제적방법시진행수마상궤적특정화,즉장설비상관적RGB안색치전환성설비무관적CIELab색도치。근거수마상궤안색공간적특성,제출기우다항식회귀적자공간자괄응보상모형산법。해모형수선대RGB안색공간진행화분,재매개자공간중건립다항식회귀모형획득예측색도치급기전환정도,연후근거해전환정도여정도역치간적관계학정시부수요대예측색도치작우화보상처리。대우전환정도저우역치적자공간,장기예측색도치여실측색도치적차치의합위RGB안색치적다항식함수,획득예측차치적다항식모형,병이용색도차치대기색도치진행보상。실험표명,해방법능구유효보상불동자공간안색특성대전환정도적영향,대대제고료안색공간전환적정도,만족안색질량검측대안색공간전환정도적요구。
Device-dependent color space is the primary problem in digital camera based printed color quality detection. Camera colorimetric characterization is used to solve the problem, which transforms device-dependent RGB data of the camera to device-independent colorimetric CIELab values. According to the color space characteristic of digital camera, sub-space compensation model using adaptive polynomial regression is proposed. Firstly, RGB color space is divided into subspaces, and polynomial regression models are created in each subspace to predict the colorimetric values and their conversion accuracy. Then the compensation and optimization for the predicted colorimetric values is carried out according to the relationship between the conversion accuracy and the pre-set threshold. If the average conversion accuracy of one sub-space is lower than the threshold, a new polynomial function is created to describe the relationship between RGB data and the colorimetric errors of the predicted colorimetric value and the measured colorimetric value, and the predicted colorimetric values then are optimized with the calculated errors. Experiments show that the method can effectively compensate the influence of different subspace color characteristic on the conversion accuracy, which greatly improve the accuracy of the color space conversion, and meet the accuracy requirement of color space conversion in color quality detection.