| With the increasing popularity of information technology,the application of biometric identification as a security authentication method for information systems is widely used.There are already relatively mature biometric identification technologies,such as face recognition and fingerprint recognition.However,there are still certain challenges for biometric systems that are oriented to complex and diverse usage environments and reducing social health hazards.In this paper,a contactless palm recognition method in complex background is given to support mobile devices to capture palm images and use hand shape wheel features and palm print features to complete the recognition task respectively,which makes the recognition method more convenient by reducing the restrictions on recognition scenarios while ensuring the correct recognition rate.In this paper,we use a self-designed image acquisition program to collect palm images captured by different mobile devices in unrestricted backgrounds by sharing QR codes to build a palm image library.In the pre-processing stage,a color equalization algorithm based on total reflection theory is used to eliminate the effect of image color differences under different capture devices,and a differential convolution algorithm is used to sharpen the palm images.In the study of hand shape recognition,this paper gives a hand shape recognition algorithm based on hand shape contour compression points.Firstly,the palm contour is segmented using the skin color maximum inter-class variance method,and the contour is morphologically processed to remove the burr of the palm contour,then the coordinate system is established using the palm contour center of mass,and the number of contour points with better comprehensive performance is found by the equal interval point taking method,and finally the support vector machine is used as a classifier to recognize the hand shape,and experiments are designed to compare and analyze the hand shape recognition algorithms.In the study of palm print recognition,the background replacement method is used to replace the background of the captured palm image with the complex background image recorded continuously in the room,and the palm image under the complex background of the labeled palm print region is used as the training data to train the improved regional convolutional neural network model.The palm print recognition model that maintains a high correct rate in complex backgrounds is obtained through experimental analysis.Finally,the final recognition results are given by fusing the hand shape recognition method and palm print recognition method through experiments using 2400 palm images with complex backgrounds from 120 individuals,and the experimental results are compared and analyzed to verify the effectiveness of the fusion method. |
