Research On Capture And Quality Evaluation Of Iris Image

In the field of biometrics, iris pattern recognition shows great advantages in stability and reliability compared with other kinds of biometric systems. The performance of iris recognition system depends on the quality of captured images. Therefore, it is essential to capture a high-quality image to improve the efficiency and accuracy of iris recognition system. The iris image acquisition device and quality evaluation algorithm are studied further to guarantee the quality of iris images for the follow-up recognition process in this paper. The contributions are as follows:(1) Design and implementation of the iris image capture devices. On the one hand, this paper analyses the effect of necessary components including the illuminant, image sensor and lens on the quality of iris image in order to find the appropriate components. On the other hand, this paper analyses the relevant factors from the acquisition mode and power supply mode. From the results, it is obvious that different acquisition modes would cause different effects on human eyes. Thus, the effect of the environment should be eliminated when the acquisition mode is chosen. In addition, an appropriate power supply mode should be chosen in order to improve the stability of acquisition system. According to the above results, an advanced iris acquisition system is designed in this paper and the characteristics are as follows: This paper adopts an acquisition pattern with non-contact and fixed focus to solve the health problem in the process of acquisition and reduce the cost. An effective feedback way based on distance is put forward to improve equipment usability and reduce the users’ compatibility. A method to remove the natural light is proposed to get rid of the influence of natural light under the non-closed environment. A voltage regulator circuit is put forward to solve the problem of uneven brightness caused by instability of illuminant in the process of capture. This paper proposes a light compensation circuit to prevent the change of pupil size in different situations.(2) Research on the Algorithm of iris image quality assessment. This paper proposes a framework for quality evaluation algorithm combining fine and rough. For the rough evaluation, image definition and iris visibility should be calculated to identify unqualified iris images. It improves the efficiency of the system but also guarantees the accuracy of positioning in the subsequent fine evaluation. Quality measures using multi-Fusion evaluation method is utilized in fine evaluation. A Iris Rotation quality measure is proposed and combines with traditional quality measures to obtain results by SVM. The analysis experiments of accuracy and predictability for the algorithm on JLU-IRIS_V3 database show that the proposed method has better performance on accuracy and predictability compared with traditional multi-measures fusion evaluation methods.In summary, on the one hand, this paper analyses the factors affecting image quality from the devices itself and the natural environment, and designs and implements a more comprehensive iris acquisition device. On the other hand, this paper defines a Iris Rotation quality measure and combines Iris Rotation with the traditional quality measures by SVM based on the analysis of the disadvantages of traditional image quality evaluation algorithm.