Study On Palmprint Recognition Algorithm Based On Multi-wavelet

Nowadays, the development of Information Technology promotes Social progressrapidly. At the same time, modern society continues to higher requirements forInformation Technology. The whole society is becoming networking and informationbecause of the development of computer, meanwhile, the network and informationsociety demands the security of information systems higher and higher. Biometrictechnology is gradually replacing passwords and keys because of its reliability and uniquewith the growing needs of security, and it is playing an increasingly important role inpeople's lives.Compared with other biometric technology, palmprint recognition has many uniqueadvantages as the biometric key member of the family. These advantages include thefollowing: Palm is unique and life‐long invariance, the image is difficult to be imitated, andthe technology itself has a high degree of user acceptance. The main features of palmprintimage are stable and clear and it is not easy to be disturbed by noise when extracting thefeature. The quality of the palmprint image is not easy affected by collection due to injuryor wear and tear affects, the possibility of information theft is much less than fingerprintsand etc. Therefore, palmprint recognition technology has been more and more concernedby academia in recent years, and the increasingly scholars are working to study thistechnology.Based on the characteristics of hand palm line image, this paper studied many keytechnologies of the palmprint identification deeply, proposed many new algorithms inaspects of location of the hand palm lines regional, processing of image rotation andidentification of feature matching. The research works included the following:1．Extraction of the separation point feature of the palm imageThe images are usually vulnerable to the impact of lighting under laboratoryconditions, feature extraction of the separation point is often fail when using the commoncharacteristics of corner detection in some conditions. In response to these problems, thispaper proposed the separation point detection algorithm of the palm based on the basetemplate and the feature template. The method identified and extracted the separationfeature points of the palm image through two template detection, achieved a desiredeffect, improved the characteristics of the image on the light palm dependence, andenhanced the efficiency of extraction of the separation point of the feature image.2．Multi‐resolution analysis of the hand palm line feature imageFull use of the multi‐resolution features of the palmprint image, the palm featureimage processed the multi‐wavelet decomposition, extracted the sub‐regions which associated with the hand palm in the processing of extraction of the palm line features.This paper used an appropriate method of dealing with these sub‐regions, completed ofthe initial match of the palmprint identification and obtained a satisfactory result.3．Polar processing of the palmprint feature texture offsetIn the process of collecting palm image, the hand position is usually not fixed and thepalm image produced rotation easily, which affected the extraction of the hand palm linefeature image. This paper used a kind of polar normalization method, solved the problemabove in theory and algorithms and reduced the influence of the image offset.4．Second match deal of the palmprint features based on neural networkThis paper used the neural network algorithm to identify the hand palm lines,matched the hand palm line features with the Hopfield network and completed thesecond recognition process of the hand palm line features based on neural network. Thequantitative data of the proximity of the palm line was also get in this processing.This paper improved several key steps of the palmprint identification, increased theaccuracy of the palmprint feature matching and improved the stability of the processing ofthe palmprint recognition, optimized the overall performance of the palmprintidentification system.