Heterogeneous Iris One-to-One Certification with Universal Sensors based On Quality Fuzzy Inference and Multi-Feature Fusion Lightweight Neural Network.

Liu Shuai,Wang Chaoqun,Wu Zukang,Liu Yuanning,Cui Jingwei,Zhu Xiaodong,Huo Guang,Li Xinlong

doi:10.3390/s20061785

Liu Shuai, Wang Chaoqun + Show 6 more

Open Access

https://doi.org/10.3390/s20061785

Copy DOI

Abstract

Due to the unsteady morphology of heterogeneous irises generated by a variety of different devices and environments, the traditional processing methods of statistical learning or cognitive learning for a single iris source are not effective. Traditional iris recognition divides the whole process into several statistically guided steps, which cannot solve the problem of correlation between various links. The existing iris data set size and situational classification constraints make it difficult to meet the requirements of learning methods under a single deep learning framework. Therefore, aiming at a one-to-one iris certification scenario, this paper proposes a heterogeneous iris one-to-one certification method with universal sensors based on quality fuzzy inference and a multi-feature entropy fusion lightweight neural network. The method is divided into an evaluation module and a certification module. The evaluation module can be used by different devices to design a quality fuzzy concept inference system and an iris quality knowledge concept construction mechanism, transform human logical cognition concepts into digital concepts, and select appropriate concepts to determine iris quality according to different iris quality requirements and get a recognizable iris. The certification module is a lightweight neural network based on statistical learning ideas and a multi-source feature fusion mechanism. The information entropy of the iris feature label was used to set the iris entropy feature category label and design certification module functions according to the category label to obtain the certification module result. As the requirements for the number and quality of irises changes, the category labels in the certification module function were dynamically adjusted using a feedback learning mechanism. This paper uses iris data collected from three different sensors in the JLU (Jilin University) iris library. The experimental results prove that for the lightweight multi-state irises, the abovementioned problems are ameliorated to a certain extent by this method.

Highlights

This paper takes lightweight one-to-one certification of multi-state iris in the same environment as the research object, and proposes a one-to-one certification method with universal sensors for Sensors 2019, 19, x FOR PEER REVIEW heterogeneous irises based on quality fuzzy inference and multi-feature entropy fusion lightweight neural networks
When evaluating the quality of the test iris, suitable concepts are selected according to the iris recognition requirements, and a quality inference machine is used to determine whether the test iris can be used for iris recognition
Because this paper focuses on lightweight iris recognition, even if we apply the multi-range setting of category labels, different iris categories can be well distinguished

Summary

Introduction

This paper takes lightweight one-to-one certification of multi-state iris in the same environment as the research object, and proposes a one-to-one certification method with universal sensors for Sensors 2019, 19, x FOR PEER REVIEW heterogeneous irises based on quality fuzzy inference and multi-feature entropy fusion lightweight neural networks. The feedback mechanism the overall process to dynamically adjust the concept of the iris in the fuzzy system and the category labels in the recognition function as the. Be predicted, which causes certain defocusing, deflection, shadowing and other problems The prerequisites for this method are listed as follows: The The dimensions of thestatus captured images areenvironment. There are a variety of use scenarios, which are mainly divided into several situations: In the iris collection state, the main settings are the collection status (collection posture, collection distance) and the external environment (illumination), which can be divided into four categories: Unconstrained state in the same environment: Iris acquisition is performed based on a lack of restriction of the acquisition posture of the acquisition target person, and the external environment is not changed for iris acquisition; Constrained state in the same environment: Iris acquisition is performed based on the restriction of the acquisition posture of the acquisition target person, and the external environment is not changed for iris acquisition; Constrained state with environmental change: Iris acquisition is performed based on the restriction of the acquisition posture of the acquisition target person, and the external environment is changed for iris acquisition; Unconstrained state with environmental change: Iris acquisition is performed based on a lack of restriction of the acquisition posture of the acquisition target person, and the external environment is changed for iris acquisition.

Methods

Results

Conclusion