Edible oil adulteration is a main concern for consumers. This paper presents a study on the use of smartphone, coupled with image processing and chemometrics, to quantify adulterant levels in extra virgin olive oil. A sequence of light with varying colours is generated on the phone screen, which is used to illuminate oil samples. Videos are recorded to capture the colour changes on sample surface and are subsequently converted into spectral data for analysis. To evaluate the performance of this video approach, partial least squares regression models constructed from such video data as well as near-infrared, ultraviolet–visible and digital imaging data are compared in the task of quantifying the level of vegetable oil in extra virgin olive oil in the range 5%–50% (v/v). The results show that the video approach (R² = 0.98 and RMSE = 0.02) yields comparable performance to baseline spectroscopy techniques and outperforms computer vision system approach. Since the smartphone-based sensor system is low-cost and easy to operate, it has high potential to become a consumer-oriented solution for detecting edible oil adulteration.

食用油掺假是消费者主要关注的问题。本文介绍了有关使用智能手机，图像处理和化学计量学来定量特级初榨橄榄油中掺假水平的研究。手机屏幕上会生成一系列颜色各异的光，用于照亮油样。录制视频以捕获样品表面的颜色变化，然后将其转换为光谱数据以进行分析。为了评估这种视频方法的性能，比较了从此类视频数据以及近红外，紫外可见和数字成像数据构建的偏最小二乘回归模型，以量化特级初榨橄榄油中的植物油含量。范围为5％–50％（v / v）。结果表明，视频方法（R ² = 0.98，RMSE = 0.02）可获得与基线光谱技术相当的性能，并且优于计算机视觉系统方法。由于基于智能手机的传感器系统成本低廉且易于操作，因此它有很大的潜力成为用于检测食用油掺假的面向消费者的解决方案。