Abstract: Camera distortion calibration plays an indispensable role in the research of robot vision. A nonlinear inverse perspective transform based method was proposed. In the light of linear perspective model and distortion model, a nonlinear inverse perspective model as well as the relation between inverse perspective transform coefficients and distortion coefficients was derived. A function with distortion coefficients as the optimum was put forward and optimized. Because only few distortion coefficients are optimized, the optimization dimension and complexity are largely reduced. During optimization, the search direction is determined by improved Newton method and a try and error linear search method is carried out along the search direction. Both the simulation and experiments verify the effectiveness and high accuracy of the method.