In this paper, an algorithm for measurement preprocessing in nonlinear target tracking is proposed by using the nonlinear measurement equation for the Kalman filter. Measurement preprocessing approach is based on the maximum-likelihood algorithm to derive estimates of the position, velocity and acceleration, and reduce the uncertainty of measurements to improve tracking performance. We consider that the original measurements from the polar co-ordinates are transformed to the Cartesian co-ordinates, and the linearization of the transformed measurements are employed in the preprocessing scheme. We incorporate the preprocessing algorithm into the variable dimension filter (VDF). We also develop a fuzzy rule-based VDF. Simulations illustrate that the measurement preprocessing-based VDF performs better than the fuzzy rule-based VDF and the VDF. (we taking into account the smaller bearing error of measurement.)