This study proposes an algorithm for predicting objects' reflectance based on the color digital camera raw data under multi-light source. The color digital camera's sensitivity functions and each light source's spectrum are measured, and the basic linear equation linking the camera raw data under multi-light source and object reflectance, including the system noise term is developed. The training color card (X-Rite 140 color card) is employed to estimate the system noise, and the multi-light source Wiener estimation algorithm is used to predict the reflectance. Two different brands of professional digital cameras are used to evaluate the algorithm. The findings demonstrate that the algorithm has the best performance under D65, U30, and HZ light sources. The Canon and Nikon cameras exhibit higher prediction accuracy for the Munsell color chart compared with the spectral reflectance obtained by the spectrophotometer, and the average CIELAB color difference is 1.59 and 1.40, respectively. It has crucial applications value in the fields of noncontact color measurement and color reproduction.