We recover the spectral reflectance of the oil painting ‘Daisy’. (a) and (b) The captured images under the fluorescent light and the tungsten light. (c), (d) and (e) The reflectance estimated and measured at selected areas (P, Q and R) on the painting. (f) and (g) The rendered image under CIE D65 and IllA.

The spectral recovery and identification of fruits. Before the experiment, a database was created by including measured reflectance of ten different kinds of fruits. Next, we captured fruit under the fluorescent light (a) and the tungsten light (b) by Canon 60D. (c) The measured and estimated reflectance of each fruit. (d) The rendering of fruits under CIE IllA. The artifacts in green were due to the shadows in the captured images. (e) All fruits were identified correctly except the pumpkin (P2) and mango (P3), due to the very similarity of their reflectance. On the other hand, while being similar in color in (a) and (b), the squash (P5) was able to be distinguished from the pumpkin (P2) and mango (P3). The identification of fruits by the spectral information is robust to the scene illuminant and to the distortion of the food shape during the processing of food. In addition, if the spectral images of the fruits at different times (g when they were ripe or rotten) are available, the identification results can be useful as a quality control measure.

We recover the spectral reflectance of outdoor materials under daylight. Based on the recovered reflectance, classifications of materials are made. (a) and (b) The captured pictures under daylight at 7:41 a.m. and 10:05 a.m. (c) The daylight spectra (normalized to be 1 at 560 nm). (d) The measured and estimated reflectance of the aloe, ceramic (the specular side), copper, ceramic (the diffuse side), gray plastic, soil, green plastic, and aluminum. (e) The rendering of the scene under CIE IllA. (f) All materials were classified (indicated by different colors). The ColorChecker Passport (CCP) was included to characterize the lighting.

Smartphones and other mobile platforms are gradually changing how we operate our daily life by integrating both the eyes (the camera) and brain (the computing unit) on a cell phone. While a handheld spectrophotometer can be used to measure the spectral reflectance of materials, the use of cell phone cameras has the benefit of lower cost and acquisition of the scene reflectance rather than of a small uniform area on a surface. We did an experiment to evaluate the spectral recovery performance of smartphones, Nokia N900, by validating CCDC. Moreover, the picture under one (rather than two) lighting condition was used to recover the scene reflectance. The validation of CCDC using the camera phone, Nokia N900, under the tungsten light only. (a) The estimated and measured reflectance of certain patches in CCDC. The numbers on the top of each plot are the spectral RMS error, and color difference under CIE D65 and IllA. The patch index (\#) is shown as well. A close spectral and colorimetric match could be achieved generally between the ground truth and our results except for patches of reddish colors. (b) Noise analysis was performed by calculating $ ho$ to tell which reflectance is likely to be predicted better under the tungsten light. Two patches were selected with small and large $ ho$. A smaller $ ho$ means a better spectral match.