Due to the difficulties in eliminating manufacturing and assembly errors, different degrees of nonlinear distortions occur in the optical system of digital video cameras, which seriously affects the accuracy of real-time image stitching. Through the comparison of a traditional distortion model of digital camera and the Brown model, we propose to utilize the simplified Brown model as the distortion model of digital video cameras. At the meantime, least squares adjustment is applied to solve the nonlinear distortion parameters based on single-photo space resection. Finally, three types of digital video cameras are calibrated by three- dimensional control field. Experimental results show that the calibration accuracy of a video camera via the proposed distortion model is distinctively higher than the classic distortionmodel, which is less than 0.5 pixel. Moreover, the calibrated parameters are stable and faithful, which effectively amends the distortion and helps to recover the relative geometric relationship inside an image.