This paper describes a method to characterizing the digital camera. The nonlinear relationship between the RGB signals generated by a digital camera and original image CIEXYZ values was obtained using the polynomial regression procedures. The reasonable structures of the polynomial were found for two digital cameras. The better number of polynomial terms was 19, yielding a modeling accuracy typically averaging 2.1~2.2 E ∆ units and maximally 9.5~10.9 E ∆ units. The experiments results showed that the polynomial regression could be used to characterize commonly digital camera.