Research Of Identification Algorithm Based On Knuckleprint And Palmprint

The recognition of the main ridge line in the hand is a relatively new type of biometric technology.The direction and shape of the main ridge line are mainly controlled by genes.It has a strong uniqueness.and stability,and it does not require high quality of the imaging picture,and is easy to identify quickly.1.Since the extraction of the hand lines is vulnerable to the uneven color of the hand in the background area,the top hat transformation is used for the hand image preprocessing for the first time.The top hat transformation can effectively remove the effect of uneven color in the background area of the hand image.Make the ridge line extraction easier,and has the characteristics of strong anti-noise ability,good reliability and so on.2.Based on the disadvantages of easy translation and rotation of the hand image collection,large area and long matching time,this paper proposes an algorithm based on the contour of the hand,including the knuckle pattern and palmprint image segmentation algorithm.The algorithm is not subject to hand translation.With the influence of the rotation,the knuckle area and palmprint area of each finger of each finger are segmented.The segmentation algorithm makes the matching area smaller and the matching search time shorter.Based on the segmentation algorithm of this paper,the distance features between finger length and knuckle are also obtained.These features make feature fusion and matching more abundant.3.Based on the shortcomings of most feature extraction algorithms such as long time consumption and noise interference in feature extraction process,this paper proposes an image feature extraction algorithm based on bidirectional vertical projection.The algorithm performs two vertical projections perpendicular to each other on the segmented finger knuckle images and the palmprint image of the fingers.The bidirectional vertical projection algorithm is simple,the feature extraction speed is fast,and the anti-interference ability is strong.Compared with other feature extraction algorithms,the feature extraction accuracy of the algorithm is high.4.The hand features extracted include the length of fingers,the distance between knuckles,knuckles,and the characteristics of the main lines of the palm lines.On the basis of single feature biometrics,in order to improve the large database in the seareh time and match correctly The performance of rate is designed to design a hierarchical search matcher with multi-features fusion time ranging from fast to slow and matching precision from low to high.The implementation method is to measure finger length,distance between knuckles and knuckles.The high efficiency of recognition is combined with the high accuracy of palmprint recognition.Select "string" rule information fusion method,first identify the finger length and the distance between the fingers of the knuckles,then identify the knuckle features,and finally identify the features of the palmprint.Thus,a multi-modal biometric identification system with both high recognition accuracy and ideal large library retrieval time performance is realized.