Palm vein main direction feature extraction method based on direction feature

Abstract

The invention discloses a method for extracting palm vein main direction features based on direction features, which is characterized in that for palm vein images, Gaussian_iRadon (G-iR) transformation is used to extract the direction feature matrix of palm vein images to obtain image main direction features, so The Gaussian_iRadon (G-iR) transformation described above to extract the directional feature matrix of the palm vein image is to use the weighting rules of the Gaussian function to the center point when determining the main direction Do(x0, y0) of the center pixel point to weaken the pixel points in the neighborhood. The contribution to the central pixel further weakens the resolution ability of the direction information extracted by the iRadon transform. The present invention adopts G-iR transform, has strengthened the association of palm vein image among different individuals, is conducive to constructing sub-library and enhancing the correlation between images in the same sub-library; at the same time, based on the better resolution ability of iRadon transform, it can be maximized as much as possible. The difference of palm vein images in different sub-databases achieves the purpose of classification and fast retrieval.

Description

technical field [0001] The invention relates to a palm vein feature recognition technology, in particular to a palm vein main direction feature extraction method based on direction features. Background technique [0002] Palm vein recognition is based on the consistency judgment of veins. Because of its good properties such as living body collection, difficult to forge, and rich and unique feature information, it has become one of the emerging biometric recognition technologies in the past decade and has attracted extensive attention from the industry. With the continuous expansion of the application field of palm vein recognition technology, the scale of palm vein database is also expanding, and the key issues that follow are the recognition accuracy and speed of large palm vein database. [0003] The palm pulse recognition algorithm can be divided into two types in terms of operation mode, namely verification mode (verification) and identification mode (identification). V...

AI Technical Summary

Problems solved by technology

Obviously, this exhaustive method is difficult to meet the real-time requirements of the identification system, and the guided retrieval method of classification or layering is an effective method to improve retrieval efficiency.

Images