The Research On Bimodal Fusion Feature Recognition Based On Palmprint And Palm Vein

With the rapid development of information technology,the issue of information security has become increasingly prominent.The biometrics technology as a highly advantageous solution fully demonstrates its importance.In order to improve the accuracy and reliability of biometric identification,the multimodal biometrics based on information fusion theory has gradually replaced the single-modality biometric technology has become a new research breakthrough in the field of identity verification and has high academic research value and practical application prospects.This paper selects the hand biometric features of the palm and palm veins for fusion recognition research.The work done is as follows:This paper builds a non-contact dual-mode image acquisition device,which can obtain palm and palm vein two kinds of biometric information at one time.A dual-mode small image database was established using this device for subsequent algorithm verification.Then,a series of preprocessing such as region of interest extraction,grayscale normalization,image denoising and image enhancement were performed on the original palmprint and palm vein images,which laid a good foundation for feature extraction and fusion recognition.This paper proposes a palmprint feature extraction method based on the origin static moment.An effective threshold segmentation and refinement algorithm is used to obtain the image of the palmprint mainline skeleton,the palmprint skeleton image is divided into multiple non-overlapping sub-blocks to extract the average origin static moments respectively and combined into the final palmprint feature vector.This paper also proposes a palm vein feature extraction method based on the global Gist feature.The enhanced palm vein image is abstracted as a scene image,and the global Gist feature of each block is extracted after being divided into blocks using the “block extraction” concept,and all the Gist features of the block are cascaded to form the palm vein feature vector.The principal component analysis method is used to reduce the dimensionality of high-dimension palm vein feature vectors so that it can be effectively used for identification.At last,this paper proposes a two-modal fusion method that uses the canonical correlation analysis method to fuse the palmprint origin feature vectors and palm veins global Gist feature vectors based on the idea of information fusion,and in the traditional CCA method to increase the category information to expand its scope of application,the use of self-built image database to verify the improved fusion method.The experimental results prove the correctness and effectiveness of the bimodal feature layer fusion recognition algorithm in this paper.The recognition rate is 96.67%,and the two kinds of single modal feature recognition are improved.