Research On Spectral Sensitivity Estimation Based On Digital Camera

Accurate prediction of spectral sensitivity of digital cameras is a basic research in the field of color science,which is crucial to other aspects of color science,such as color constancy,light source estimation,color correction,etc.The object imaging quality of digital camera is mainly affected by two factors.On the one hand,the imaging quality of the camera is affected by external factors,such as the spectral reflectance of the light source and the scene target surface;On the other hand,it is affected by the processing algorithm of color signal response value inside the camera.In fact,the research on spectral sensitivity estimation is to improve the ability of camera sensors to process color signal response values through analysis and modeling,so as to promote the development of color science.The research of camera spectral sensitivity is not only applied in color science,but also widely used in the field of computer vision.In the field of computer vision,object imaging is the premise and foundation of all scientific research.Only by extracting and analyzing the pixel information and color information of the object imaging can a series of studies on subsequent image features be carried out.such as edge detection,corner extraction,etc.The purpose of spectral sensitivity estimation research is to improve the quality of the object imaging,so it is of great significance to the research of spectral sensitivity estimation.In the paper,through the research on the camera imaging process and fully considering the influence of various factors in the camera imaging process,a camera spectral sensitivity estimation algorithm is proposed.The main innovations are as follows.1 Based on the analysis of camera imaging principle and considering the characteristics of camera sensor itself,the camera imaging process is modeled by quadratic programming method,and smoothness constraint is added on the basis of quadratic programming to improve the stability and robustness of the estimation algorithm.2 The conversion of the camera imaging process into a mathematical model will cause illposed problems,and the noise of the camera itself,the estimation errors of the light source data and the spectral reflectance data of the scene target surface will affect the accuracy and robustness of the camera spectral sensitivity estimation algorithm.Considering the above factors,the spectral sensitivity estimation problem of the camera is transformed into a multiobjective optimization algorithm.3 The most ideal result of camera imaging is that the color of the surface of the object imaged by the camera is the same as that measured by CIE standard observers,that is,the color of the scene target is highly restored.Considering that the camera sensor only responds to the color signal entering the camera and cannot control the color signal entering the camera.Therefore,a method of pre-filtering the camera is proposed to control the color signal entering the camera lens to improve the imaging quality.By solving the transmittance of the camera prefilter,each camera can obtain a unified imaging result.