Feature Fusion And Identification Of Infrared Hand And Its Residual Trace Image

Compared with traditional password authentication,biometric feature is not easy to lose and has uniqueness and good stability.At present,most biometric identifications use visible images,and face some technical bottlenecks in information acquisition,feature extraction and recognition,which limit the application and development of the technology.Focusing on the problem of how to improve the robustness and accuracy of the feature recognition system,this paper attempts to implement a recognition system using the infrared hand and its residual trace(handprint)images.Infrared thermal imaging technology was used to acquire the heat information of hand and handprint,the original images were preprocessed with grayscale,smoothing,binarization and contour extraction to obtain the distinct contours.According to the different types of hand image feature and handprint image feature,different feature extraction methods were adopted,including comer extraction,ellipse fitting,ellipse secondary fitting,etc.,and the eigenvector space was established.Multimodal feature fusion and dimension reduction method were used to conduct feature fusion and dimension reduction of extracted hand features and handprint features,and appropriate features were selected for recognition.The one-against-one method was used to implement multiclass recognition algorithm of support vector machine.In this paper,an individual reeognition method based on infrared hand and handprint image features fusion is implemented.More than 1200 infrared hand and handprint images of 12 people are acquired for experiments,and a high success accuracy of individual identification(70?90%)was obtained.The experimental results verify that it is feasible to use infrared hand and handprint images for individual recognition,and feature fusion and feature dimensionality reduction are better than using single biometric feature recognition.The method proposed can effectively improve the robustness and accuracy of the recognition system.