Objective To construct a spectral measurement system based on digital imaging equipment, and to achieve "fast, accurate, low-cost" spectral measurement results. Methods Three-channel RGB values were obtained from each color with the color table ECI2002 by a digital camera. Different weights were assigned according to the similarity of samples and the cosine similarity were introduced, to reconstruct the spectral reflectance which could represent the original color information. Meanwhile, lens distortion correction, light uniformity correction and system repetitiveness correction were proposed for the color measurement system. Results The proposed method brought the average color difference to below 0.62, RMS to below 0.01 and controlled the computation time within 1 min when the number of training sample was 10.Conclusion The digital imaging measurement system constructed in this paper could enhance the measurement accuracy and improve the efficiency, save the time, and therefore has engineering application potential.