KR20020060271A - Realtime pupil detecting method for iris recognition - Google Patents

Abstract

PURPOSE: A method for detecting pupils of eyes is provided to perform an accurate iris recognition which is insensitive to an influence of an illumination for reflecting an eyeball image for realizing an iris detection having an enhanced calculating amount and velocity. CONSTITUTION: In an iris recognizing system including an illumination, a video camera for obtaining an eyeball image of a human body, and a PC having a frame grabber, a basic point is searched firstly by an eyeball image reduced as a predetermined rate for decreasing a calculating amount of a photographed actual eyeball image(S1). A basic point in an eye pupil is detected from the reduced eyeball image(S2), and boundary reserved areas between an iris and an eye pupil of the reduced eyeball image is decided(S3). A radius and a center of a circle most near to a boundary reserved area are calculated using more than three boundary reserved points distanced most out of the boundary reserved areas, a position and a size of an eye pupil are decided, and an eye pupil area is detected(S4).

Description

Real time pupil detecting method for iris recognition

The present invention relates to iris recognition, and in particular, real-time for iris recognition to detect a pupil from the boundary of the iris and the pupil in the human eye image for iris recognition having a stable and fast personal identification in the human body part. It relates to a pupil detection method.

Recently, there has been a growing interest in stable and fast personal identification methods such as fingerprint recognition and iris recognition, and a highly reliable personal identification system is an essential industrial application in security of restricted areas and credit card security. It can be used in many ways.

One of the highly individualized parts of the human body is the iris. The probabilistic features of the iris have considerable freedom and sufficient uniqueness to identify an individual. Individuals can be identified using probabilistic independence on the features extracted from the pattern inside the iris. Pupil detection is a process that must be preceded in iris recognition, and fast pupil detection is essential for real time iris recognition.

In general, the conventional pupil detection method proposed by Daugman of the existing iris recognition system obtains circular projection at all positions of the image, and then uses Gaussian Convolution for the derivatives of the projection. After estimating the boundary line at the largest value, the method finds the pupil from the human eye image by finding the position where the circular boundary component is the strongest using the estimated boundary value.

However, the conventional method has a disadvantage in that a large amount of time is required for pupil detection due to an increase in the amount of calculation due to the projection of the entire iris and the derivative of the obtained projection. In order to obtain an iris picture, illumination is required. In the conventional method, when the light source is positioned in the pupil, there is a problem of inaccurate pupil boundary caused by the influence of the light source.

In addition, in the case of the iris recognition system using the conventional method, a search is performed on a portion excluding the iris region that includes reflected light or may have a lighting effect by setting the region excluding the illumination in the iris region as the iris region. There is a problem in that the accuracy is lowered.

The present invention has been proposed to solve the above problems of the prior art, and an object of the present invention is the effect of illumination on the eye image for iris detection with improved throughput and speed compared to the existing iris recognition method for personal identification. The present invention provides a real-time pupil detection method for iris recognition insensitive to high accuracy.

1 is a flowchart illustrating a real-time pupil detection method for iris recognition according to the present invention.

2 is a system configuration of connecting the iris recognition software of a computer by illuminating the eyeball image of the human body with a camera under the influence of lighting.

3 is a two-dimensional and three-dimensional representation of a Gaussian template for reference point detection to find a pupil from a photographed eye image.

4 is an eyeball image of a search result for a reference point using a Gaussian template.

5 is a diagram showing a direction for obtaining a pupil boundary candidate from a reference point;

Fig. 6 shows the pupil detection results of eyeball images under the influence of illumination.

Explanation of symbols on the main parts of the drawings

1: lighting 2: video camera

3: PC with frame grabber installed

In order to achieve the above object, the present invention provides an iris recognition system having an illumination (1), a video camera (2) for acquiring an eyeball image of a human body, and a PC (3) with a frame grabber installed: a real eye photographed First searching the reference point with the eyeball image reduced in a predetermined ratio to reduce the amount of computation of the image (S1); Detecting an intrapupillary reference point from the reduced eye image (S2) and determining a boundary candidate point between the iris and the pupil of the reduced eye image (S3); And using the three or more boundary candidate points far away from the home among the boundary candidate points of the iris and the pupil to determine the position and size of the pupil by detecting the radius and center of the circle closest to the boundary candidate (S4). It provides a real-time pupil detection method for iris recognition, characterized in that consisting of steps.

Hereinafter, exemplary embodiments of the present invention will be described in detail with reference to the accompanying drawings.

1 is a flowchart illustrating a real-time pupil detection method for iris recognition according to the present invention. FIG. 2 shows a system in which an eye image of a human body is illuminated by a camera under the influence of lighting, and connected to an iris recognition software of a computer. The system includes a lighting 1, a video camera 2, and a PC 3 having a frame grabber installed therein. . The PC 3 is programmed with Visual C ++ using iris recognition software.

In order to reduce the amount of computation for the captured eye image by introducing a mean sampling technique, the candidate region is first searched from the reduced eye image as follows (step S1).

N: horizontal equal ratio, M: vertical equal ratio, i: 0 to (N-1), j: 0 to (M-1)

For example, when the pupil is detected using the eyeball image reduced to one quarter of the actually photographed eyeball image (N = 2, M = 2), it is expressed as Equation 2 below.

Geometric deviations of the lighting components formed on the reduced eye image are obtained, their average values are calculated, and modeled as Gaussian waveforms as shown in the following equation (for example, = 1.2).

Where x is the horizontal position of the template, y is the vertical position of the template, and σ is the size of the filter.

As shown in FIG. 3, the reference point is detected by performing template matching so that the reference point is selected in the pupil of the eye image using the modeled template (step S2).

In the eyeball image, the illumination inside the pupil is the only part where the rapid gray-level change occurs, and thus, it is possible to extract a stable reference point as shown in FIG. 4.

As shown in FIG. 5, a profile, which is a linear waveform in eight directions, is extracted from the reference point as shown in FIG. 5. In order to detect a boundary in N one-dimensional signals, a boundary candidate mask h (n) corresponding to a gradient is generated, and convolutions of the profile () and the boundary candidate mask (h (n)) are generated. Using the convolution, the boundary candidate waveform X (n) is generated.

For example, in the case of Laplacian of Gaussian Mask, a boundary candidate mask is displayed as follows.

Using a general edge detection algorithm such as zero-crossing on a waveform obtained by convolution, the boundary between the pupil and the iris is found and determined as the boundary candidate point (step S3).

Using three or more points (C4, C5, C6) farthest from the boundary candidate points, the radius and center coordinates of the circle representing the pupil closest to the boundary candidate are obtained. The error is the sum of the difference between the distance from the center to the points and the radius, and a pair is obtained by determining a pair that minimizes the error and by using the circle, as shown in FIG. (Step S4).

The following Table 1 compares the speed of detecting pupils from the eyeball image with the application method of Iris and the algorithm proposed in the present invention.

Pupil Detection Technique Detection speed (Based on 486DX66) Iris pupil detection 408 ms Pupil Detection Technique of the Invention 150 ms

Table 2 below shows the results of experiments on 50 ocular images obtained under normal conditions.

accuracy 100% Average speed (Pentium II-350) 8 ms

Therefore, the reduced eye image is extracted from the digital eye image taken from the camera for iris recognition used for personal identification, and the iris pupil detection speed is improved through the detection using the reference point and the boundary candidate points of the iris and the pupil. Can improve.

As described above, the real-time pupil detection method for iris recognition according to the present invention extracts a reduced eye image, not a search for the entire area of the human eye image, and detects the reference point and the boundary candidate points between the iris and the pupil. It is possible to reduce the amount of computation and to increase the pupil detection speed for iris recognition.In addition, it is possible to use the full iris region instead of the limited iris region due to the light source effect in the existing iris recognition system. There is an effect that can be improved.

Although the above has been described with reference to a preferred embodiment of the present invention, those skilled in the art will be able to variously modify and change the present invention without departing from the spirit and scope of the invention as set forth in the claims below. It will be appreciated.

Claims (3)

In an iris recognition system having an illumination (1), a video camera (2) for acquiring an eyeball image of a human body, and a PC (3) equipped with a frame grabber: First searching a reference point with the eyeball image reduced in a predetermined ratio to reduce the amount of calculation of the actual eyeball image (S1); Detecting an intrapupillary reference point from the reduced eye image (S2) and determining a boundary candidate point between the iris and the pupil of the reduced eye image (S3); And Detecting the pupil area by determining the position and size of the pupil by obtaining the radius and center coordinates of the circle closest to the boundary candidate using three or more boundary candidate points far from home among the boundary candidate points of the iris and the pupil (S4) Real time pupil detection method for iris recognition, characterized in that consisting of. The method of claim 1, Extracting a reduced eye image from the digital eye image taken from the video camera 2 for iris recognition used for personal identification and detecting a circular pupil area using a reference point and a boundary candidate point of the iris and the pupil Real-time pupil detection method for iris recognition, characterized in that. The method of claim 1, A pupil detection method for iris recognition, or a real-time pupil detection method for iris recognition, characterized by using an effect of illumination in detecting a reference point thereof.