A characteristic method of camera response function(CRF) based on PCA

Response function is one of the important characteristics of CCD and CMOS. Traditional measure method of response function needs to fulfill strict measurement conditions, and has the shortcomings that low measurement efficiency and high measurement cost. This paper uses PCA method to compute 5 eigenvectors corresponding to 5 maximum eigenvalues of the response function space based on 201 existing response functions. Expressing the existing response functions by the linear combination of the 5 eigenvectors, a rather accurate result was achieved. 4 type of digital cameras' response functions which were not included in the 201 existing response functions were expressed by the eigenvectors of camera response functions' space. The result is rather accurate which proves the effectiveness of the method.