Method for Extracting Orientation Field of Markov Random Field Fingerprint Based on Prior Knowledge

Abstract

The invention discloses a method for extracting a fingerprint direction field under a Morkov random field based on priori knowledge. The problems of incapability of accurately extracting fingerprint direction field information in a low-quality area and high generation rate of pseudo-singular point information in a model method in the prior art are mainly solved. The method is implemented by the following steps of: (1) inputting a fingerprint image, extracting the priori knowledge of a direction field of the fingerprint image, obtaining a block angle matrix of the fingerprint image by using a gradient method, and obtaining the probabilities of block angles by using a statistical method; (2) fitting the priori knowledge of the direction field, the obtained block angle matrix and the probabilities of the block angles, and constructing a priori-knowledge-based Morkov random field model of the fingerprint image; and (3) extracting a final direction field of the fingerprint image by using the constructed model. The priori knowledge of the fingerprint direction field is fitted on the basis of the conventional Morkov random field, and a new fingerprint direction field extraction model is constructed, so that the matching and retrieval performance of the fingerprint image are improved, and the method can be used for extracting the direction field of a low-quality fingerprint image.

Description

technical field [0001] The invention belongs to the technical field of image processing, relates to the preprocessing of fingerprints and the extraction of direction fields, and can be used for the extraction of direction fields of low-quality fingerprint images. Background technique [0002] The fingerprint direction field is a very important information of the fingerprint, which shows the global information of the fingerprint, and describes the basic shape, structure and ridge direction information of the fingerprint macroscopically. In the actual application system, the correct estimation of the fingerprint direction field is beneficial to other fingerprint processing, such as fingerprint enhancement, fingerprint segmentation, and fingerprint singular point detection, so the fingerprint direction field detection method with excellent performance can improve the performance of other fingerprint processing processes . The existing fingerprint direction field extraction met...

AI Technical Summary

Problems solved by technology

[0003] The gradient-based direction field estimation method is a traditional method using gradients to estimate the direction field. This method is based on the average of the squared gradient vectors of the fingerprint ridges, which not only doubles the gradient vector angle, but also squares the length of the gradient vector. That is, the gradient vector is used as the complex number of the square. This method has low computational complexity and can quickly calculate the direction field of the fingerprint image, and can correctly extract the fingerprint direction field on high-quality fingerprint images, but it is easily affected by noise. , low robustness, poor effect on low-quality fingerprint image extraction

[0004] The model-based direction field extraction method uses mathematical principles to construct a basic fingerprint ridge information model, and then uses the fingerprint image database to continuously improve the model parameters to achieve the optimal effect. The direction field obtained by this type of algorithm is relatively smooth. This method It is easy to generate false singular point information, which reduces the performance of other processing processes such as fingerprint image matching, and it is difficult to correctly extract the direction field of low-quality fingerprint images

[0005] The direction field estimation method based on the filter response is to use the filter to estimate the response to multiple directions, for example, the variance of the fingerprint ridge in 16 directions, and the direction with the largest response is the direction of the ridge. This method is susceptible to noise. The influence of low robustness, the effect of extracting low-quality fingerprint images is very poor

Images