Full convolutional neural network-based fingerprint sweat pore extraction method

Abstract

The invention relates to a full convolutional neural network-based fingerprint sweat pore extraction method. The method includes the following steps that: step 1, a high-definition fingerprint image is acquired, sweat pore positions and ridge positions in each fingerprint image are marked, data augmentation is performed on the marked fingerprint images, so that a marked data set can be obtained; step 2, a full convolutional neural network model is constructed, initial parameters and a loss function are set, the marked data set is adopted to train the full convolutional neural network model, so that a trained full convolutional neural network model is obtained; step 3, the trained full convolutional neural network model is adopted to predict the preliminary region probability of the sweat pores and ridges of a test fingerprint image; and step 4, pseudo sweat pore regions are removed from preliminary sweat pore regions according to the characteristics of the sweat pores, and therefore, real sweat pore regions and center coordinates can be obtained. According to the full convolutional neural network-based fingerprint sweat pore extraction method of the invention, the full convolutional neural network is adopted to learn and extract the characteristics of sweat pores of different shapes and different sizes, and therefore, the accuracy of sweat pore extraction can be improved.

Description

technical field [0001] The invention relates to the field of fingerprint identification, in particular to a fingerprint sweat pore extraction method based on a fully convolutional neural network. Background technique [0002] Because of the uniqueness and permanence of fingerprints, fingerprint features are widely used in personal identification as the most commonly used biometric features; the current automatic fingerprint recognition system (AFRS) generally uses the minutiae features of fingerprints for fingerprint recognition, although the current fingerprint The identification system (AFRS) has a good accuracy rate, but with the continuous improvement of the public's personal security needs, the automatic fingerprint identification system (AFRS) needs to use more fingerprint features to improve its accuracy rate. The third-level features of fingerprints, like the minutiae features of fingerprints, are proven to be unique and permanent; [0003] Fingerprint sweat pore ex...

AI Technical Summary

Problems solved by technology

[0004] In order to overcome the problems of various sweat pore feature shapes and inaccurate detection results in the existing sweat pore extraction technology, the present invention proposes a fingerprint sweat pore extraction method based on a fully convolutional neural network, which uses a fully convolutional neural network to learn Extract sweat pore features of different shapes and sizes, thereby improving the accuracy of sweat pore extraction

Images