Camera filter design based on optimized basis functions

Abstract

Color cameras are widely used in many fields such as printing industry, graphic arts, medical treatment, and environment. On the premise of saving cost, in order to ensure that the color rendering effect of the camera is as close as possible to the imaging of the human eye, a method of using color filters to correct the total spectral response curve of the color camera is proposed. The principle of correction is to make the total spectral response of the system meet the Luther condition. By adding such a filter, the adjusted camera sensitivity function can be very close to a certain linear transformation of the color matching function of the human visual system. Due to the manufacturing process, the transmittance of the produced filter can only be a smooth curve. Starting from the factors that affect the accuracy of the filter simulation, we express the transmittance of the filter as a certain smoothness in the calculation process combination of basis functions. Different basis functions will lead to different results. Here we use discrete cosine transform basis functions, polynomial basis functions, Fourier basis functions and radial basis functions to conduct experiments. Under the condition of each basis function, a corresponding optimal spectral transmittance curve will be obtained. Taking the 14 standard test colors recommended by the International Commission of Illumination as a reference, the CIE1976 color difference formula is used to calculate the theoretical color difference of the corrected camera under the condition of different basis functions. Finally, the performance of the basis function is evaluated from three indicators: Vora-Value, NRMSE and color difference.