In this paper, a Backpropagation (BP) neural network classifier based on the Optical-Electrical Hybrid Power Spectrum Analysis System is presented. In this system, to realize the real time processing of cell image, the CCD camerra, high resolution CRT and LCLV is applied to complement the input image. In considerattion of the complexity, variety, high noise and non-parameter distribution of the cell pattern,the BP neural network is served as cell pattern classifier. In our experiment, the correct recognition rate is above 90%.