Design considerations for intelligent iris image acquisition terminal under restricted conditions

O Introduction

The iris is a visible ring-shaped tissue located between the black pupil and the white sclera on the surface of the human eye. Under near-infrared light of a certain frequency, it can present rich texture information, such as spots, stripes, fine lines, coronal shapes, crypts and other physiological details. The visible biological tissue structure of the human iris depends on the initial conditions of the mesoderm in the embryonic period of infants. Its distribution in the population may be random or chaotic, but it is stable throughout life after birth, and the iris texture of each person is absolutely different. Statistics show that the iris texture has hundreds of degrees of freedom. Even for the same genotype, the phenotypic expression of the iris is unrelated. Because the unique texture image of the human iris is suitable for automatic identity recognition, it has the characteristics of high efficiency, accuracy and non-replicability. The basic principle is mainly to determine the identity of the human body by comparing the similarity between the iris texture image features. The core is to use a large number of multiple algorithms on the computer, and use pattern recognition, image processing and other methods to describe and match the iris texture features of the human eye (such as the early application of Gabor wavelet to encode iris texture and use Hanmin distance to match iris templates), so as to achieve automatic human identity identification.

1 Composition of iris recognition system

The iris recognition system consists of a software system and a hardware system. The software system is the iris information processing system, which is used to realize functions such as iris image processing, user registration, user identification, iris image storage management, and iris feature storage management. The block diagram is shown in Figure 1. The hardware system includes the iris image acquisition system and the hardware environment that supports the operation of the iris information processing system.

The more mature computer core algorithms on the market include Daugman technology from Cambridge University, MJRLIN technology from Bath University, and technology from the Chinese Academy of Sciences. The experimental system in this paper uses MIRLIN technology from Bath University.

2 Analysis of iris texture image acquisition technology under restricted conditions

The International Eye Safety Standards Association requires that the iris image acquisition device be a real-time, automatic, and non-invasive iris optical imaging device. At present, the mainstream iris lens in the world uses a CMOS progressive scanning camera with a resolution of 640×480 pixels or more, which requires at least 100 pixel acquisition points within the diameter of the human iris to retain the characteristics of the original iris image as much as possible. According to the ISO/IEC 19794-6 image standard and to ensure the need for shooting in a dark environment, the light source is a near-infrared light source with a wavelength of 720 to 900 nm and an irradiation power of <0.5 mW/cm2, which is harmless to the eyes. The transmission rate of the camera is set to 251 frames/s to ensure the real-time transmission of the video stream. During the shooting process, the central processing module performs real-time analysis on each frame of the image until the frame meets the discrimination standard, and the data frame is transmitted to the iris recognition processor core module for corresponding discrimination processing. The overall hardware design architecture is shown in Figure 2.

In a total of 352,730 experiments, we found that several important limiting factors for the success of iris image acquisition in the actual system include human-computer interaction, light intensity and other system external limitations. Therefore, the applicability design of the system is very important. In order to achieve fully automatic intelligent iris acquisition, through tens of thousands of experiments, we concluded that the system design has the following considerations:

2.1 Distance Sensing

First, the camera module must be equipped with an external user distance sensor to monitor the presence of an individual. The sensor can be contact type (such as a card reader or password switch module) or contactless module (such as an infrared sensor or shoulder probe module). Considering the future compatibility of the system, the interface design with the CPU is recommended to use GPIO connection.

[page]

2.2 Human-computer interaction

The built-in application software of the camera can guide users to perform corresponding operations in multiple languages. The built-in distance sensing sensor module will feed back the user's distance to the central processor, and then guide the user to adjust the head position until entering the appropriate imaging area in the form of voice output. Alternatively, the use of visual multi-color indicator lights can also achieve this effect.

2.3 Adjustable imaging depth

The imaging depth of the camera is another limiting factor. To reduce the impact of this limiting factor and improve the performance of the system, the camera should have a certain range of imaging depth. Within this range, the optical lens can automatically adjust the focal length for real-time imaging. The imaging depth of this experimental system can be normally imaged within the range of 22 to 36 cm. This reduces the restrictions on user operations.

2.4 Head tilt detection

The camera also needs to have a "head tilt detection" function module to detect whether the user's head tilts beyond a certain range, which makes it impossible to perform subsequent matching judgments normally. This eliminates the limitation of authentication failures caused by improper head position and ensures the stability of the system.

2.5 Glasses/Sunglasses Testing

The lens of glasses will reflect the incident flash of the camera, resulting in a strong reflection spot in the collected iris image. If the bright spot is located in the iris area, it will cause matching errors. Therefore, the camera needs to have a special polarization performance to eliminate reflections, make the colors more vivid, enhance the image contrast and extract image features. Figure 3 is a comparison of the iris imaging effects collected by wearing glasses in the experiment without polarization and after polarization.

2.6 Height adjustable

Another limiting factor of the iris acquisition system is the height difference of different users. This system uses a slide industrial design to allow the camera to be raised and lowered (as shown in Figure 4), thereby achieving universality.

2.7 Other limiting factors

All iris acquisition methods in this study are static or semi-static, that is, the user is required to remain relatively still during the iris acquisition process. This poses certain technical difficulties for dynamic image acquisition of the human body in motion. The United States has relatively mature technology for real-time iris image acquisition during action, but the product price is relatively expensive and its widespread application needs to be improved. Only by adopting advanced technology and greatly reducing hardware costs can it be widely used in my country.

This paper analyzes several limiting factors encountered in the actual application of iris recognition systems and proposes solutions. It has broad guiding significance for the industrial design and automation development of existing iris recognition terminal systems. There are many types of biometric technologies, and the application prospects of iris recognition technology, which is gradually being applied, are very obvious. China is currently in a leading position in the world in the manufacturing strength of electronic and information products. Chinese biometric technology manufacturers are also catching up with the world's advanced in the production of iris recognition products. The 2007 Information Security Technology "Technical Requirements for Iris Recognition Systems" national standard was issued and implemented, providing a strong guarantee for the rapid promotion and application of iris recognition technology in the field of identity authentication in my country. Due to its unique characteristics, iris recognition technology will play a vital role in solving human identity authentication in China, a country with one-tenth of the world's population, and its application prospects are immeasurable.