A Large-Scale Palmprint Encoding and Comparison Method

Abstract

The invention relates to a large-scale palmprint encoding and comparison method, comprising: performing image rotation and translation on a large-scale registered palmprint image and a palmprint image to be recognized according to a unified coordinate system; The size is cut into several palmprint image blocks, wherein the palmprint image block cut out from the registered palmprint image is called the registered palmprint image block, and the palmprint image block cut out from the palmprint image to be recognized is called the palmprint image to be recognized block; construct a feature matrix and a detail point matrix for each palmprint image block; construct a palmprint fast screening tree; use the feature matrix and detail point matrix of the palmprint image block to make all the palmprint image blocks to be identified in turn in the fast screening tree Retrieve the palmprint image blocks that have similar feature evaluation indicators with the palmprint image blocks to be recognized, and use the relationship between the palmprint image blocks and the palmprint images to select the most similar registered palmprint image blocks to be recognized. The palmprint image is used as the finally recognized palmprint image.

Description

technical field [0001] The invention provides a large-scale palmprint coding and comparison method, which belongs to the field of biological feature recognition, and specifically relates to a palmprint recognition method. Background technique [0002] With the continuous advancement of computer technology and information processing and identification technology, palmprint recognition technology has developed rapidly and has become a mature and acceptable biometric identification technology, which can be used in public security, network management, social security, Electronic access control and other widely used technologies in many fields have important theoretical research significance and market application value. [0003] Palmprint recognition technology is divided into two processes: (1) palmprint feature extraction process. By reading the palmprint image, the palmprint area in the palmprint image is found, and the features in the palmprint area that can distinguish the...

AI Technical Summary

Problems solved by technology

[0005] In the large-scale palmprint recognition scenario, the above methods are not suitable: (1) In the process of palmprint feature extraction, only using the above local and overall features is not enough to accurately distinguish the palmprint features of large-scale palmprint images, which leads to Recognition accuracy drops significantly, so that the so-called "big library accuracy attenuation" effect occurs, and the present invention then additionally constructs stable features for distinguishing large-scale palmprints, thereby significantly improving recognition accuracy; (2) the main stream disclosed in patents and documents The palmprint recognition methods all use the idea of ​​"one-by-one comparison" between the palmprint image to be recognized and the registered palmprint image, so it takes a lot of time

Images