颗粒图像的颜色校准与表征-中国粉体技术

2024年30卷第4期

HTML (0)

颗粒图像的颜色校准与表征
Color calibration and characterization of particle images

黄作杰¹ ，周骛¹ ，徐喜庆² ，裴昌蓉² ，蔡天意¹ ，蔡小舒¹

1. 上海理工大学能源与动力工程学院，上海市动力工程多相流动与传热重点实验室，上海 200093；2. 中国石油大庆油田有限责任公司勘探开发研究院，多资源协同陆相页岩油绿色开采全国重点实验室，黑龙江大庆 163712

引用格式：

黄作杰，周骛，徐喜庆，等. 颗粒图像的颜色校准与表征［J］.中国粉体技术，2024，30（4）：104-114.HUANG Z J， ZHOU W， XU X Q， et al. Color calibration and characterization of particle images［J］.China Powder Science andTechnology，2024，30（4）：104−114.

DOI：10.13732/j.issn.1008-5548.2024.04.010

收稿日期：2024-01-30，修回日期：2024-05-04，上线日期：2024-06-22。

基金项目：国家自然科学基金项目，编号：52376163；上海市科学技术委员会启明星培育（扬帆）计划项目，编号：22YF1429600。

第一作者简介：黄作杰（1997—），男，硕士生，研究方向为图像法颗粒测量。E-mail：stevenerv@foxmail. com。

通信作者简介：周骛（1986—），女，教授，博士，博士生导师，研究方向为颗粒与两相流测量。E-mail：zhouwu@usst. edu. cn。

摘要：【目的】减少图像法测量系统中颗粒成像的颜色失真，并对颗粒进行颜色表征。【方法】采用色卡对颗粒成像系统进行标定，使用色彩校正算法建立实际拍摄的色卡的颜色值与D65光源下的色卡理论颜色值之间的映射关系，以此对颗粒图像进行色彩校正，对比分析基于多项式回归的 6种色彩校正算法，对基于校正前、后色卡的平均色差和算法的曝光适应性进行算法测试，并在白色LED灯光与偏黄的卤素灯光照射下分别验证色彩校正效果。【结果】三阶多项式色彩校正算法的回归精度最高，在白色LED光源照射下，校正前、后色卡的24个色块的平均色差由38. 67下降到3. 82，但测试发现，三阶多项式色彩校正算法不具备曝光适应性；三阶根多项式色彩校正算法在回归精度上接近三阶多项式色彩校正算法且具有良好的曝光适应性。【结论】基于色卡标定和色彩校正算法可以在一定程度下减少系统的颗粒成像的色偏，在6种校正算法中三阶根多项式色彩校正算法能够提高回归精度和曝光适应性；针对颗粒系，可结合颗粒的平均色品坐标和颗粒数量进行表征，针对单个颗粒，可以使用颜色矩和主要颜色进行表征。

关键词：图像法；颗粒测量；色彩校正；颜色表征

Abstract

Objective Particle color is an important parameter across various sectors， reflecting the composition， purity， and quality of particles. Different particles colors may also have different physical and chemical properties. At present， image-based particle characterization mainly focuses on particle size and shape， and the characterization of particle color has not been systematically studied. The national standard GB/T 38879-2020 Color Image Analysis for Particle Size Analysis and the international standard ISO/PWI TS 19673 Particle characterization - Color image analysis， which is led by China， recognize that particle color is another important parameter in particle image-based analysis besides particle size and shape. However， the spectral distribution of the light source in imaging system， the absorption of lens to the light， and the spectral response of the camera sensor will affect the color properties of the captured particle images. Therefore， it is necessary to reduce the influence caused by the above factors through color correction. Additionally， the paper addressed how to quantitatively and intuitively characterize the color and distribution of particles and particle groups.

Methods In this paper， a particle color measurement device was built， using a color card to calibrate the color of the device.We compared and verified the correction effects of 6 common color correction algorithms under the illumination of white LED and halogen lamps. The device was also used to capture particle images from drug capsules， and the color correction algorithm was used to correct the color of the particle images. The color information of particles was extracted and characterized by processing the captured particle images.

Results and Discussion Through the verification of the 6 common polynomial color correction algorithms， it was found that the linear color correction algorithm with white balance constraint had lower regression accuracy. After correction， the average color difference of the 24 color blocks in the color card was reduced from 38. 67 to 13. 61 under the illumination of white LED light source， and reduced from 43. 13 to 21. 52 under the illumination of halogen lamps. This color correction algorithm maintained good white balance. The third-order polynomial color correction algorithm had the highest regression accuracy. Under the illumination of white LED light source， the average color difference of 24 color blocks in the color card was reduced from 38. 67 to 3. 82 after correction， and the average color difference was reduced from 43. 13 to 3. 92 under the illumination of halogen lamps.The third-order root polynomial color correction algorithm ranked the second. Under the illumination of white LED light source，the average color difference of 24 color blocks in the color card was reduced from 38. 67 to 4. 15 after correction by using this color correction algorithm， and from 43. 13 to 4. 24 after correction under the illumination of halogen lamps. Experiments showed that， unlike the third-order polynomial color correction algorithm， the third-order root polynomial color correction algorithm demonstrated good exposure invariance with no color deviation across different exposure intensities.

Conclusion In this paper， the color card was used to calibrate the imaging system under two kinds of light sources， and 6 kinds of commonly used color correction algorithms were used to correct the captured particle images. The experimental results showed that using color cards to calibrate the imaging system could effectively reduce the color bias in the imaging system， thereby theoretically aligning the particle color more closely to the RGB values under the D65 standard light source. Among the 6 color correction algorithms tested in this paper， the third-order root polynomial color correction algorithm showed good performance in terms of correction efficiency and exposure invariance， making it the recommended choice for color correction. For the characterization of particle colors， this paper transformed the average RGB values of particles to their corresponding chromaticity coordinates. The particle count was used to describe the relationship between the particle color and count in the particle system. Color moments and main colors were used to represent the color characteristics of individual particles， among which， the main color was the area where the color of particles was most concentrated in the Lab color space. This approach is beneficial in mitigating extraction errors of the main color caused by uniform color values appearing locally in the particle image.

Keywords：imaging method； particle measurement； color correction； color representation

参考文献（References）

［1］GALATAD L， MESZAROSL A， KALLAI-SZABO N， et al. Applications of machine vision in pharmaceutical technology： a review［J］.European Journal of Pharmaceutical Sciences，2021，159：105717.

［2］GOSSELIN R， DURÃO P， ABATZOGLOU N， et al. Monitoring the concentration of flowing pharmaceutical powders in a tableting feed frame［J］.Pharmaceutical Development and Technology，2017，22（6）：699-705.

［3］LIU Y， ZHAN L， WEN L， et al. Effects of particle size and color on photocuring performance of Si3N4 ceramic slurry by stereolithography［J］.Journal of the European Ceramic Society，2021，41（4）：2386-2394.

［4］GAUCI A， DEIDUN A， MONTEBELLO J， et al. Automating the characterisation of beach microplastics through the application of image analyses［J］.Ocean & Coastal Management，2019，182：104950.

［5］COWGER W， GRAY A， CHRISTIANSEN S H， et al. Critical review of processing and classification techniques for images and spectra in microplastic research［J］.Applied Spectroscopy，2020，74（9）：989-1010.

［6］SUNOJ S， IGATHINATHANE C， SALIENDRA N， et al. Color calibration of digital images for agriculture and other applications［J］.ISPRS Journal of Photogrammetry and Remote Sensing，2018，146：221-234.

［7］褚璇，王伟，张楠楠，等. 基于颜色特征的含黄曲霉毒素玉米颗粒的检出方法［J］.中国粮油学报，2015，30（4）：112-118.CHU X， WANG W， ZHANG N N， et al. Detection method of corn kernels infected aflatoxin based on color features［J］.Journal of the Chinese Cereals and Oils Association，2015，30（4）：112-118.

［8］HUNT R W G， POINTER M R. Measuring colour［M］.Hoboken： John Wiley & Sons，2011：25-30.

［9］SHARMAG， BALAR. Digital color imaging handbook［M］.Boca Raton： CRC Press，2017：76-89.

［10］葛超. 数字图像颜色校正方法研究［D］.武汉：武汉大学，2022.GE C. Research on color correction methods of digital image［D］.Wuhan： Wuhan University，2022.

［11］ASHRAF M， SAPAICO L R. Evaluation of color correction methods for printed surfaces［C］//2018 Colour and Visual Computing Symposium （CVCS）.New York： IEEE，2018：1-6.

［12］COLANTONI P， THOMAS J B， HARDEBERG J Y. High‐end colorimetric display characterization using an adaptive training set［J］.Journal of the Society for Information Display，2011，19（8）：520-530.

［13］GOÑI S M， SALVADORI V O. Color measurement： comparison of colorimeter vs. computer vision system［J］.Journal of Food Measurement and Characterization，2017，11：538-547.

［14］MA K， SHI J. Colorimetric characterization of digital camera based on RBF neural network［C］//Optoelectronic Imaging and Multimedia Technology VII. California： SPIE，2020，11550：282-288.

［15］ASTER R C， BORCHERS B， THURBER C H. Parameter estimation and inverse problems［M］.Massachusetts： Elsevier，2018：55-63.

［16］HONG G， LUO M R， RHODES P A. A study of digital camera colorimetric characterization based on polynomial modeling［J］.Color Research & Application，2001，26（1）：76-84.

［17］FINLAYSON G D， MACKIEWICZ M， HURLBERT A. Color correction using root-polynomial regression［J］.IEEE Transactions on Image Processing，2015，24（5）：1460-1470.

［18］FANG F， GONG H， MACKIEWICZ M， et al. Colour correction toolbox［C］// Proceedings of 13th AIC Congress 2017.Jeju： Korea Society of Color Studies，2017：48-53.

［19］QUINTANA J， GARCIA R， NEUMANN L. A novel method for color correction in epiluminescence microscopy［J］.Computerized Medical Imaging and Graphics，2011，35（7/8）：646-652.

［20］International Organization for Standardization. ISO 12640-2：2004， Graphic technology-Prepress digital data exchange-Part 2： XYZ/sRGB encoded standard colour image data （XYZ/SCID）［S/OL］.［2024-05-04］.http： www. iso. org/obp/ui/#iso：std：iso：12640：-2：ed-1：v1：en.

［21］STRICKER M A， ORENGO M. Similarity of color images［C］// Storage and retrieval for image and video databases III.California： SPIE，1995：381-392.

颗粒图像的颜色校准与表征Color calibration and characterization of particle images

颗粒图像的颜色校准与表征
Color calibration and characterization of particle images