评估巧克力饼干的颜色：仪器比色法、数字图像分析和计算机视觉相对于人类感知的比较性能
Evaluating the Color of Chocolate Chip Cookies: Comparative Performance of Instrumental Colorimetry, Digital Image Analysis, and Computer Vision Relative to Human Perception.
摘要(中文译文)
食品颜色通常使用色度计进行量化；然而，这些仪器在家庭或小规模生产环境中可能不实用。更易获取的替代方法，如数字图像分析和计算机视觉，可能在传统仪器不可用时提供可靠的颜色量化。本研究的目的是确定最能准确、可靠地预测消费者感知曲奇颜色的颜色测量方法和采样策略。研究比较了三种表征巧克力曲奇颜色的方法，并评估其与人类感知的一致性：仪器色度法（HunterLab色度计）、数字图像分析（Adobe Photoshop）和计算机视觉分析（OpenCV/Python）。100名消费者使用设计为近似L*、a*和b*颜色构念的15 cm线段量表，对15个曲奇样品（3块曲奇×5个商业品牌）的感知亮度、红色度和黄色度进行了评价。仪器测量和基于图像的测量在两种采样条件下采集：（1）每块曲奇上四个标准化表面位置（选定点）和（2）全表面平均（整块曲奇）。在各方法之间，选定点采样通常比整块曲奇采样产生更高的L*值。多变量分析表明，所有方法均能区分不同品牌。然而，与感官构型的一致性在选定点采样下的仪器色度法和整块曲奇采样下的计算机视觉中最强。预测建模进一步表明，选定点仪器色度法最能一致地预测感知亮度和黄色度，而整块曲奇计算机视觉对感知红色度产生了表现最佳的预测。总之，在本数据集中，选定点色度法与感知曲奇颜色表现出最密切的对应关系，而计算机视觉作为资源有限环境中颜色量化的实用替代方法展现出强大前景。
Abstract
Food color is commonly quantified using colorimeters; however, these instruments may be impractical in home-based or small-scale production settings. More accessible alternatives, such as digital image analysis and computer vision, may offer reliable color quantification when conventional instrumentation is unavailable. The objective of this study was to identify the color-measurement method and sampling strategy that most accurately and reliably predict consumers' perceived cookie color. Three approaches to characterizing the color of chocolate chip cookies were compared and evaluated for concordance with human perception: instrumental colorimetry (HunterLab colorimeter), digital image analysis (Adobe Photoshop), and computer-vision analysis (OpenCV/Python). One hundred consumers evaluated the perceived lightness, redness, and yellowness of 15 cookie samples (3 cookies × 5 commercial brands) using 15-cm line scales designed to approximate L*, a*, and b* color constructs. Instrumental and image-based measurements were collected under two sampling conditions: (1) four standardized surface locations on each cookie (selected point) and (2) whole-surface averaging (whole cookie). Across methods, selected-point sampling generally produced higher L* values than whole-cookie sampling. Multivariate analyses showed that all methods differentiated among brands. However, agreement with sensory configurations was strongest for instrumental colorimetry under selected-point sampling and for computer vision under whole-cookie sampling. Predictive modeling further indicated that selected-point instrumental colorimetry most consistently predicted perceived lightness and yellowness, whereas whole-cookie computer-vision generated the best-performing predictions of perceived redness. In conclusion, selected-point colorimetry showed the closest correspondence with perceived cookie color in this dataset, while computer vision demonstrated strong promise as a practical alternative for color quantification in resource-limited settings.
如何引用
AMA
Natalia Calderón, Eniola Ola, Han-Seok Seo. Evaluating the Color of Chocolate Chip Cookies: Comparative Performance of Instrumental Colorimetry, Digital Image Analysis, and Computer Vision Relative to Human Perception.. Journal of food science. 2026; doi:10.1111/1750-3841.71133.
APA
Natalia Calderón, Eniola Ola, Han-Seok Seo (2026). Evaluating the Color of Chocolate Chip Cookies: Comparative Performance of Instrumental Colorimetry, Digital Image Analysis, and Computer Vision Relative to Human Perception.. Journal of food science. https://doi.org/10.1111/1750-3841.71133
GB/T 7714
Natalia Calderón, Eniola Ola, Han-Seok Seo. Evaluating the Color of Chocolate Chip Cookies: Comparative Performance of Instrumental Colorimetry, Digital Image Analysis, and Computer Vision Relative to Human Perception.[J]. Journal of food science, 2026 doi:10.1111/1750-3841.71133.
在 PubMed增强版 阅读全文
翻译全文 · AI 文献问答 · PDF 原文下载 · 引用导出