Abstract: In order to improve the police investigation process middle ear grain recognition accuracy, overcome traditional algorithm in view of the high dimension extrusion ear grain deformation,this paper proposes a ear grain extrusion condition recognition method of alarm.On the basis of the traditional algorithm,the use of computer vision ear grain local mean standard deviation and image processing to deal with the ear grain image grey value,the fusion algorithm of PSA and LDA algorithm,ear grain pattern vector of extrusion condition recognition nonlinear dimension reduction processing,has a strong ability to judge the ear grain extrusion condition recognition classifier, finish high precision don't by mistake.The simulation results show that the algorithm effectively reduces the external factors such as extrusion effect on ear pattern recognition technology,improve the identification accuracy.