Abstract
Camera spectral sensitivity plays an important role in color correction, color constrancy and color science. In this paper, we propose a method to estimate the camera spectral sensitivity function using Quadratic Programming. The optoelectronic conversion function (OECF) of the camera determines that the RGB response of the camera is linear, so the spectral sensitivity of the camera can be obtained through linear regression. In fact, the camera has certain noise, and linear regression will be affected by the noise, which is also an important and difficult point in estimating spectral sensitivity. In addition, linear regression is not stable. Different light sources and spectral reflectance (spectral reflectance of scene object) will have a great impact on spectral sensitivity estimation. In order to solve the above problems, smoothness constraints and curve characteristics (positive and unimodal) of spectral sensitivity are added to the linear regression. Through simulation experiments, it was found that the estimated value obtained by the algorithm of estimating spectral sensitivity by using the quadratic programming problem based on smoothness constraints is very close to the real value and has very high stability.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call