A pattern recognition model for haze identification with atmospheric backscatter lidars is proposed. The process of building the characteristics sample database for haze pattern recognition is described in detail. The classification of haze particles by using Bayesian discriminant function, as the selection basis of haze optical characteristics vector, is presented. Computer simulation for the proposed pattern recognition model of haze identification is carried out. Two self-calibration approaches are employed to check the validity and stability of the model. By analyzing the applicability of this model for atmospheric lidars, the advantage of polarized high spectral resolution lidar (HSRL) is highlighted.