An algorithm of spectral reflectance function reconstruction without sample training can integrate prior information

Abstract

In the case of known lighting conditions and environmental factors, how to reconstruct the spectral reflectance function of the object from the camera response value is the key technology of color detection in industrial production. In order to solve the problem that the direct reconstruction method is difficult to integrate into the prior information and is affected by the noise, and the operation of the sample training method is complicated, an algorithm of spectral reflectance function reconstruction without sample training can integrate prior information input is designed. The algorithm uses the camera response value and the reasonable estimate of the spectral reflectance function as input. By cyclic iteration, the spectral reflectance function of the object is reconstructed step by step. Using this method and pseudo inverse method for simulation experiments, the experimental results show that the proposed algorithm can well combine the prior information with the camera response value, and under the same conditions, the algorithm reduces the mean RMSE of the pseudo inverse method by about 32%, and has high accuracy, easy operation and strong compatibility specialty.