We present a lens distortion model based on the Gaussian function. The model is a potential source function of the popular Brown distortion model and requires the practical estimation of one Boolean and one real parameter. We also present a general technique for lens distortion identification, which consists of three steps: image data acquisition, localization and indexation/matching of image features, and the estimation of distortion parameters. The method uses a structured light technique, in which bar patterns are indexed by a Gray code. Acquired images are automatically processed using a multistep approach that localizes and indexes calibration points. An iterative analysis of differences between localizations of distorted points and their undistorted counterparts is proposed to estimate distortion model parameters. To compute undistorted localizations, the method estimates a homography matrix that is based on both undistorted data and iterative processing of distorted coordinates, which improves compensation accuracy. Experiments with three cameras show that the indexing strategy significantly decreases compensation error (from 0.26 to 0.09 pixels). The newly introduced Gaussian model is shown to be slightly more accurate and considerably more stable than the popular Brown model.