The main theme of these processes is to identify the person of the palm of given data base. Orientation feature has been demonstrated to be one of the most promising features for palm print recognition. The consistent orientation pattern (COP) hashing method to accelerate search as much as fast to extract the feature the steerable filter is used. Based on that features only, the consistency pattern can be identify the result. In existing system, has the feature extraction using Gabor filter and the classification is performed using different classifier (SVM, NN- classifier). Proposed a method of combining two-dimensional Gabor transforms and invariant moments to extract palm print feature, and using multilayer towards feedback neural network for training palm print images to recognize This method first pretreated the collected palm print images and got the region of interest (ROI), then constructed a set of Gabor filters to get ROI eigenvectors, combined with the palm images’ invariant moment features together as the input of the neural network to train and recognize. Experiments show that the method is effective. Palm print recognition usually includes the processes of palm image acquisition, preprocessing, feature extraction, classification and identification. Palm print image acquisition primarily completes tasks of acquisition and preservation the original images, and then through the pre-processing stage to carry out the operation of removing noises, enhancement, segmentation, positioning and normalized, and form the standard palm database, then calculate and extract the images’ feature information in the database. Some of the palm print data from palm template database and the others form the samples to be identified.