Research On Spectral Reflectance Reconstruction Based On Multispectral Imaging System

The transmission and reproduction of color information is closely related to people’s daily life in this information age of the 21 st century.Digital image has been widely applied in people’s life. While people enjoy digital image equipment bring great convenience, While people enjoy digital image equipment bring great convenience,they are trouble enjoy digital and the original inconsistent problem in varying degrees. But the key is how to efficiently and accurately acquire the surface spectral reflectance spectrum based on the color reproduction.Spectrophotometer can directly measure the spectral reflectance of the object, but the measured area and surface evenness have restrictions with low efficiency. Scanner has higher spectral efficiency but can not be measured non-contact measurement and measurement area is finite. Obviously, they can not meet the requirements of color reproduction which including the traditional Chinese painting, oil painting art as well as other artwork based on spectrum. Multi-spectral camera can efficiently get to the spectral reflectance of the surface, but it is complex, expensive and low efficiency limits its application in practice.Digital camera image acquisition as a major tool in people’s daily life has been widely used, but a typical commercial digital camera with only three channels, can only collect the color information under certain lighting conditions, can not meet the high-precision color reproduction It needs. Therefore, how to efficiently obtain the spectral reflectance of the surface color has become a hot field of scientific research.The main component analysis and R matrix method are the most commonly used by the current color science and technology workers while rebuilding the surface spectral reflectance. Principal component analysis(PCA) is a kind of mathematical statistics, R matrix method is to put the spectrum is divided into the same color spectrum black and basic stimulus spectrum. In this dissertation, principal component analysis(PCA) and R matrix method are theoretically derived and verified by experiments. The experimental results show that the spectral reflectance through the use of matrix R reconstruction are better than that of the principal component analysis(PCA)on the spectral accuracy and chromaticity accuracy.Based on Principal Component Analysis and R matrix method in-depth study, this paper proposes a new spectral reconstruction algorithm- based matrix R reconstruction method(PPR), and design experiments to verify the feasibility and reliability of the algorithm. The experimental results show that the reconstructed spectral reflectance by the new method on precision of spectral and chromaticity accuracy are better than that of the principal component analysis(PCA), compared with the R matrix method, the spectral accuracy improved, but compared with color accuracy has little difference.Finally, color cards, oil painting, traditional Chinese painting as a test sample, using multi-spectral imaging system designed in this paper combined with the spectrum reconstruction algorithm proposed in this paper(PPR) to reconstruct the spectral reflectance of test samples, and then use the color difference and the knowledge of computer graphics to achieve a true representation of the test sample surface color and the color reproduction quality of the test sample has been made subjective evaluation. Preliminary results show that the proposed spectrum reconstruction algorithm(PPR) has a high value and practical significance. In the field of color reproduction, color quality assessment and cultural heritage and other digital reproduction it will have a broad application prospect.