WO2015103953A1

WO2015103953A1 - Method and device for extracting iris image image under condition of non-uniform illumination

Info

Publication number: WO2015103953A1
Application number: PCT/CN2015/070058
Authority: WO
Inventors: 孙贤军; 董记平; 夏鹏
Original assignee: 上海帝仪科技有限公司
Priority date: 2014-01-09
Filing date: 2015-01-04
Publication date: 2015-07-16
Also published as: CN104778729A

Abstract

Disclosed are a method and device for extracting an iris image under the condition of non-uniform illumination. The method comprises: filtering a non-uniform illumination intensity component from an image containing an iris image (S102); calculating a colour difference in the filtered image (S104); and conducting iris image extraction according to the colour difference (S106). In the embodiments, by filtering a non-uniform illumination intensity component, the influence of the condition of non-uniform illumination on the iris image extraction can be avoided.

Description

Iris extraction method and device under uneven illumination

Technical field

Embodiments of the present invention generally relate to the field of image processing, and more particularly to an iris extraction method and apparatus under uneven illumination conditions.

Background technique

In today's information age, how to accurately identify a person's identity and protect information security is a key social issue that must be resolved. To this end, biometric identification technology has quietly emerged and has become a frontier research topic in the world of information security management.

Biometric identification technology refers to the use of the physiological characteristics or behavioral characteristics inherent in the human body for personal identification. The iris identification technology is an important branch of biometric identification technology. For one person, the center of his eye is a black pupil, and the iris is the annular tissue between the outer edges of the pupil, which presents interlaced texture features similar to spots, filaments, stripes, and crypts. The iris is hardly changed in one's life, and the iris of different people is completely different. Iris identification technology is the application of computer image processing technology and pattern recognition technology in the field of personal identification. Because of its high stability and high accuracy, it can also make people get rid of the cumbersome memory credit card number, bank account number, ID number, and network access number. Therefore, it is widely used in banking, public security, airport, network security and other industries. .

The typical iris identification is mainly composed of image acquisition, image preprocessing, feature coding and feature matching. In the image preprocessing process, iris extraction is especially critical, and its execution time and accuracy will directly affect the recognition speed and accuracy of the entire iris identification process. Traditional iris extraction methods include Hough transform circle detection method, Daugman's method based on differential integral operator, and boundary detection method adopted by Wildes et al. They both model the iris boundary as a circular ring with two inner and outer boundaries, and extract the iris by extracting the inner and outer circles of the ring.

However, during the actual acquisition process, the iris is usually not a complete circular ring since the upper eyelid always blocks the upper half of the iris. And the outer boundary of the iris is often It is more ambiguous, that is, the transition between the iris and the sclera (the white part of the sclera, that is, the outer part of the eyeball, which is the outermost layer of the eye) is not obvious, so it is difficult to detect the outer boundary by the above-mentioned conventional boundary detection method. In particular, it should be pointed out that the traditional method is time-consuming and laborious, and it is difficult to meet the requirements of real-time fast and accurate detection.

To this end, many researchers have also carried out related work on this non-ideal collection of iris extraction. For example, in the invention patent application CN103246871A, the inventor first uses the existing Daugman method to extract the inner boundary of the image, and then selects the rectangular regions on the left and right sides of the inner boundary such that the rectangular region covers the left and right sides outside the inner boundary as much as possible. The iris is then gray-scale transformed into the rectangular region, and the Canny operator is used to detect the boundary of the rectangular region. Specifically, when performing boundary detection on a rectangular area, a pixel value lower than the threshold is set to 0 according to a predetermined threshold, and otherwise set to 1, thereby obtaining a boundary point set. It can be seen that in the invention patent application CN103246871A, a common image processing basic operation means called threshold segmentation is used to extract the target of interest.

However, it is often difficult to robustly detect iris boundaries using only a single threshold operation. Especially for the case of binocular iris recognition, since the eyes are usually in uneven illumination, this makes a single, fixed threshold not adaptable to both eyes at the same time.

Summary of the invention

In view of the above problem that uneven illumination conditions affect iris extraction, embodiments of the present invention provide an iris extraction method and apparatus under uneven illumination conditions.

According to an aspect of the present invention, there is provided an iris extraction method under uneven illumination conditions, comprising: filtering out an uneven illumination intensity component from an image containing an iris; calculating a color difference in the filtered image; Iris extraction is performed according to the color difference.

In one embodiment, the filtering out the uneven illumination intensity component from the image containing the iris comprises: converting a color mode of the image such that the uneven illumination intensity component in the converted image and The color components are separated.

In one embodiment, the image is converted from an RGB color mode to a Lab color mode to obtain corresponding image components in the L channel, the a channel, and the b channel such that The uneven illumination intensity component and the color component in the converted image are separated, wherein the image component in the L channel includes the uneven illumination intensity component, an image component in the a channel, and the The image component in the b channel includes the color component.

In one embodiment, the calculating the color difference in the filtered image comprises calculating the color difference using an image component in the a channel and an image component in the b channel.

In one embodiment, the color difference is calculated using image components in the a channel and image components in the b channel according to the following formula:

Where ΔE _ab represents the chromatic aberration, Δ(a) represents the difference between the two values in the a channel, and Δ(b) represents the difference between the two values in the b channel.

In one embodiment, the performing iris extraction according to the color difference comprises: comparing the color difference with a predetermined color difference threshold to obtain a first comparison result; and performing iris extraction according to the first comparison result.

In one embodiment, the filtering out the uneven illumination intensity component from the image comprising the iris further comprises filtering the non-uniform illumination intensity component from the image components in the L channel.

In one embodiment, filtering the uneven illumination intensity component from the image components in the L channel comprises: performing a logarithmic operation on the image components in the L channel; the log transformed image Performing a fast Fourier transform on the component; filtering the low frequency portion of the image component after the fast Fourier transform by high-pass filtering; performing inverse transform of the high-pass filtered image component on the fast Fourier transform; The inversely transformed image component is subjected to an exponential operation.

In one embodiment, the performing iris extraction according to the color difference further comprises: comparing the color difference with a predetermined color difference threshold to obtain a first comparison result; and calculating the filtering after filtering the uneven light intensity component a gray value corresponding to the image component in the L channel; comparing the gray value with a predetermined gray threshold to obtain a second comparison result; calculating the first comparison result and the first of the second comparison result Intersection; and performing iris extraction based on the first intersection.

In one embodiment, the performing iris extraction according to the color difference further comprises: calculating a brightness of any point in the image component in the L channel after filtering the uneven light intensity component minus a point around the point a difference between the average values of the brightness of the predetermined number of points; comparing the difference with a predetermined brightness difference threshold to obtain a third comparison result; calculating the first intersection and the second of the third comparison result Intersection; and performing iris extraction based on the second intersection.

In one embodiment, the difference is calculated according to the following formula:

Where (x, y) represents the coordinates of any point in the image component in the L channel after filtering out the uneven illumination intensity component, and f(x, y) represents the point of the coordinate (x, y) Brightness value,

Represents a non-negative integer set and n represents a non-negative integer.

In an embodiment, performing iris extraction according to the second intersection includes: determining, respectively, whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if so, Deriving a foreground pixel in the same target area as the adjacent foreground pixel to obtain a plurality of target areas; respectively calculating a number of foreground pixel points owned by the plurality of target areas; and the target area having the largest number Determined to be an iris region; and perform iris extraction in the iris region.

In one embodiment, performing iris extraction in the iris region includes: eliminating holes in the iris region, non-adjacent boundaries, resulting in a complete iris region; and performing iris extraction in the intact iris region.

In one embodiment, the holes in the iris region, non-adjacent boundaries are eliminated according to the following formula:

Where A represents the iris region, α represents any point in A, B represents a collection of structural elements, and b represents any point in B.

Represents a two-dimensional integer grid.

In one embodiment, the holes in the iris region are also eliminated according to the following formula, Adjacent borders are based on:

Where A represents the iris region, α represents any point in A, and B represents a collection of structural elements.

Represents a two-dimensional integer grid.

According to another aspect of the present invention, there is provided an iris extraction apparatus under uneven illumination conditions, comprising: a filtering device for filtering out uneven illumination intensity components from an image containing an iris; and a computing device for Calculating the color difference in the filtered image; and an iris extracting means for performing iris extraction based on the color difference.

In one embodiment, the filtering device includes a conversion unit for converting a color mode of the image such that the uneven illumination intensity component and the color component in the converted image are separated.

In one embodiment, the conversion unit converts the image from an RGB color mode to a Lab color mode to obtain corresponding image components in the L channel, the a channel, and the b channel, such that in the converted image The uneven illumination intensity component and the color component are separated, wherein the image component in the L channel includes the uneven illumination intensity component, and the image component in the a channel and the image component in the b channel include The color component.

In one embodiment, the computing device includes a first computing unit for calculating the color difference using image components in the a channel and image components in the b channel.

In one embodiment, the first calculating unit calculates the color difference by using an image component in the a channel and an image component in the b channel according to the following formula:

In one embodiment, the iris extraction device includes: a first comparison unit configured to compare the color difference with a predetermined color difference threshold to obtain a first comparison result; and an iris extraction unit configured to The results were compared for iris extraction.

In one embodiment, the filtering device further comprises: a filtering unit for filtering out the uneven illumination intensity component from image components in the L channel.

In one embodiment, the filtering unit includes: a first operation module, configured to perform a logarithm operation on image components in the L channel; and a second operation module, configured to convert the logarithmically transformed image The component performs a fast Fourier transform; a third operation module is configured to filter the low frequency portion of the image component after the fast Fourier transform by high-pass filtering; and a fourth operation module, configured to filter the high-pass The image component performs an inverse transform of the fast Fourier transform; and a fifth operation module is configured to perform an exponential operation on the inverse transformed image component.

In one embodiment, the iris extraction device includes: a first comparison unit, configured to compare the color difference with a predetermined color difference threshold to obtain a first comparison result; and a second calculation unit, configured to calculate the filtering a gray value corresponding to the image component in the L channel after the uneven light intensity component; a second comparing unit, configured to compare the gray value with a predetermined gray threshold to obtain a second comparison result; a third calculating unit, configured to calculate a first intersection of the first comparison result and the second comparison result; and the iris extraction unit is configured to perform iris extraction according to the first intersection.

In one embodiment, the iris extraction device further includes: a fourth calculation unit, configured to calculate a brightness of any one of the image components in the L channel after filtering the uneven illumination intensity component, minus the a difference between the average values of the brightness of the predetermined number of points around the point; a third comparing unit for comparing the difference with a predetermined brightness difference threshold to obtain a third comparison result; the fourth calculating unit, Computing a second intersection of the first intersection and the third comparison result; and the iris extraction unit is configured to perform iris extraction according to the second intersection.

In one embodiment, the fourth calculating unit calculates the difference according to the following formula:

Represents a non-negative integer set and n represents a non-negative integer.

In an embodiment, the iris extraction device further includes: a determining unit, configured to divide Determining whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if present, dividing the foreground pixel and the adjacent foreground pixel into the same target area a plurality of target regions are obtained thereby; a fifth calculating unit, configured to separately calculate a number of foreground pixel points owned by the plurality of target regions; and a determining unit configured to determine the target region with the largest number as an iris region; And the iris extraction unit performs iris extraction in the iris region.

In one embodiment, the iris extraction device further includes: an elimination unit for eliminating holes in the iris region, non-adjacent boundaries, to obtain a complete iris region; and the iris extraction unit at the complete Iris extraction is performed in the iris region.

In one embodiment, the elimination unit eliminates holes, non-adjacent boundaries in the iris region according to the following formula:

Represents a two-dimensional integer grid.

In one embodiment, the elimination unit also eliminates holes, non-adjacent boundaries in the iris region according to the following formula:

Represents a two-dimensional integer grid.

As will be understood from the detailed description below, according to an embodiment of the present invention, after filtering out the uneven illumination intensity component, the extraction is gradually performed to avoid the influence of the uneven illumination condition on the iris extraction.

DRAWINGS

The above and other objects, features and advantages of the embodiments of the present invention will become <RTIgt; In the drawings, several embodiments of the invention are illustrated in the

1 illustrates a flow chart of an iris extraction method under uneven illumination conditions, in accordance with an embodiment of the present invention;

2 illustrates a schematic diagram of raw RGB before color mode conversion, in accordance with an embodiment of the present invention;

3 illustrates a schematic diagram of a color mode converted L channel in accordance with an embodiment of the present invention;

4 illustrates a schematic diagram of a channel after color mode conversion, in accordance with an embodiment of the present invention;

FIG. 5 illustrates a schematic diagram of a b-channel after color mode conversion in accordance with an embodiment of the present invention; FIG.

6 illustrates a schematic diagram of chromatic aberrations of a channel and b channel in accordance with an embodiment of the present invention;

FIG. 7 illustrates a schematic diagram of a left eye region color difference effect according to an embodiment of the present invention; FIG.

8 illustrates a schematic diagram of a first comparison result obtained by comparing the color difference in FIG. 7 with a predetermined color difference threshold according to an embodiment of the present invention;

9 illustrates a schematic diagram of filtering image components in an L channel after uneven illumination, in accordance with an embodiment of the present invention;

FIG. 10 illustrates a schematic diagram of a second comparison result obtained by comparing the grayscale value corresponding to FIG. 9 with a predetermined grayscale threshold value according to an embodiment of the present invention; FIG.

11 illustrates a schematic diagram of performing an operation of removing a local mean for the gray value corresponding to FIG. 9 according to an embodiment of the present invention;

12 is a diagram illustrating a third comparison result obtained by comparing a luminance difference value after the partial mean subtraction operation in FIG. 11 with a predetermined luminance difference threshold value according to an embodiment of the present invention;

Figure 13 illustrates a schematic diagram of a first intersection of the first comparison result in Figure 8 and the second comparison result in Figure 10, in accordance with an embodiment of the present invention;

14 illustrates a schematic diagram of a second intersection of the first intersection of FIG. 13 and the third comparison result of FIG. 12, in accordance with an embodiment of the present invention;

15 illustrates a schematic diagram of a plurality of target regions obtained by dividing each foreground pixel into different target regions, respectively, according to an embodiment of the present invention;

16 illustrates a schematic diagram of the iris region having the largest number of foreground pixel points in FIG. 15 in accordance with an embodiment of the present invention;

17 illustrates a schematic view of a complete iris region obtained by eliminating holes, non-adjacent boundaries in an iris region, in accordance with an embodiment of the present invention;

FIG. 18 illustrates a structural block diagram of an iris extraction apparatus under uneven illumination conditions according to an embodiment of the present invention.

In the various figures, the same or corresponding reference numerals indicate the same or corresponding parts.

detailed description

The principles and spirit of the present invention are described below with reference to a few exemplary embodiments illustrated in the drawings. It is to be understood that the examples are given only to enable those skilled in the art to understand the invention, and not to limit the scope of the invention in any way.

1 illustrates a flow chart of an iris extraction method under uneven illumination conditions, including steps S102 through S106 as follows, in accordance with an embodiment of the present invention.

Step S102, filtering out the uneven light intensity component from the image containing the iris.

In step S104, the color difference is calculated in the filtered image.

In step S106, iris extraction is performed according to the color difference.

In view of the uneven illumination condition, each point in the image has a non-uniform illumination intensity component. In the present embodiment, by filtering the uneven illumination intensity component, the influence of the uneven illumination condition on the iris extraction can be avoided.

Thus, embodiments of the present invention are capable of improving non-uniform illumination between regions within an image. However, in the related art, the luminance difference between the image frames is often used, for example, the luminances in the respective image frames are each extracted and the average value is calculated, and then the average value is used as the average luminance of each image frame. Thus, although the difference in luminance between image frames can be improved, it is not possible to improve non-uniform illumination between regions within the image.

In an embodiment, filtering the uneven light intensity component in step S102 can be This is achieved by converting the color mode of the image, for example, by converting the uneven illumination intensity component and the color component in the converted image, thereby facilitating filtering out the uneven illumination intensity component.

In this embodiment, the image can be converted from RGB color mode to Lab color mode (Lab is often used as an informal abbreviation for CIE 1976 (L*, a*, b*) color mode, ISO 11664-4: 2008 (E)/CIE S 014-4/E: 2007). Since the Lab color mode only has the uneven illumination intensity component in the image component of its L channel, the separation of the uneven illumination intensity component and the color component in the converted image is achieved.

2 to 5 respectively illustrate images before and after the above color mode conversion. Among them, in the original RGB image before the conversion illustrated in FIG. 2, it can be clearly seen that the left eye and the right eye are in uneven illumination conditions. In the image of the converted L channel illustrated in FIG. 3, uneven illumination conditions at the left and right eyes still exist. However, in the images of the converted a-channel and b-channel illustrated in FIGS. 4 and 5, respectively, there is no uneven illumination at the left and right eyes, and only the chromatic aberration exists.

It will be appreciated by those skilled in the art that the above-described color mode conversion is merely exemplary and not limiting, and any other embodiment capable of achieving a separation of uneven illumination intensity components, even without separation, can directly filter out the uneven illumination intensity. Embodiments of the components should all be included in the scope of protection of the present invention.

In one embodiment, since there is a significant color difference between the iris and the skin in the Lab color mode, iris extraction can be performed using the image component in the a channel of the Lab color mode and the color difference of the image component in the b channel. That is, the color difference is compared with a predetermined color difference threshold to obtain a first comparison result; and iris extraction is performed according to the first comparison result. Wherein, the color difference can be calculated according to the following formula and the extraction effect thereof can be illustrated in FIG. 6.

Where ΔE _ab represents the color difference, Δ(a) represents the difference between the two values in the a channel, and Δ(b) represents the difference between the two values in the b channel.

In one embodiment, the unevenness can also be filtered out from the image components in the L channel. The uniform light intensity component is calculated, and the gray value corresponding to the image component in the L channel after filtering the uneven light intensity component is calculated, and the iris extraction is performed more accurately according to the gray value and the color difference. In this embodiment, performing more precise iris extraction based on the gray value and the color difference may be achieved by comparing the gray value with a predetermined gray threshold to obtain a second comparison result; And comparing the result of the comparison with the first comparison result; and extracting the iris according to the intersection. For the purpose of making the context description of the embodiment of the present invention clearer and more consistent, the joint extraction effect according to this embodiment can be referred to FIG. As is apparent from the image illustrated in Fig. 13, most of the iris regions have been whitened and can be clearly distinguished except for the eyeglass frame.

In this embodiment, filtering the uneven illumination intensity component from the image components in the L channel can perform homomorphic filtering on the image components in the L channel, that is, performing logarithmic operations, fast Fourier transforms, High-pass filtering, inverse transform of fast Fourier transform, and exponential operation. Since the relative variation of the uneven illumination intensity in the image is small, it can be regarded as a low frequency component of the image, so that the uneven illumination intensity component can be filtered out by high-pass filtering.

Those skilled in the art will appreciate that the above-described operations involving homomorphic filtering are merely exemplary and not limiting, and any other embodiment that enables filtering of non-uniform illumination intensity components should be included in the scope of the present invention.

In addition, in the iris extraction process, since the eyeglass frame is close to the iris and generally has a similar color, the effect of iris extraction is affected. In order to solve this problem, in one embodiment of the present invention, the operation of removing the local mean may be performed, that is, calculating the brightness of any point in the image component in the L channel after filtering out the uneven illumination minus the reservation around the point. The difference between the average values of the brightness of the number of points, specifically, the difference is calculated according to the following formula:

Represents a non-negative integer set and n represents a non-negative integer. Comparing the difference with a predetermined brightness difference threshold to obtain a third comparison result; calculating a second intersection of the first intersection and the third comparison result; and performing iris extraction according to the second intersection. In this way, the influence of various objects including the eyeglass frame, such as eyelashes, on iris extraction can be removed.

This will be described in detail below with reference to FIGS. 7 to 14. For the sake of simplicity of explanation, FIGS. 7 to 14 only take the right eye as an example, and the case of the left eye is similar to the right eye.

It will be understood by those skilled in the art that before performing the operation of removing the local mean, it is possible to manually select whether or not the glasses have been worn for different images, and if the result of the selection indicates that the glasses have been worn, the first comparison described with reference to FIGS. 7 to 14 is used. The result, the second comparison result, and the iris extraction method of the third comparison result, otherwise only the iris extraction method involving the first comparison result and the second comparison result is used. Meanwhile, those skilled in the art should also understand that since the above-mentioned iris extraction method involving the first comparison result, the second comparison result, and the third comparison result is fast, it is more preferable that all of the manual selection steps are not required. The image performs the above-described iris extraction method involving the first comparison result, the second comparison result, and the third comparison result. All of the above methods should be included in the scope of protection of the present invention.

7 illustrates a schematic diagram of chromatic aberration between image components in a channel and image components in a b channel, which has been described in detail above, and thus will not be described again herein, in accordance with an embodiment of the present invention.

8 illustrates a schematic diagram of a first comparison result obtained by comparing the color difference in FIG. 7 with a predetermined color difference threshold, in which there is a spectacle frame and an enlarged iris edge region, in accordance with an embodiment of the present invention.

FIG. 9 illustrates a schematic diagram of filtering image components in an L channel after uneven illumination according to an embodiment of the present invention, and how to acquire image components in the L channel has been described in detail above, and thus will not be described herein.

FIG. 10 illustrates a schematic diagram of a second comparison result obtained by comparing the grayscale value corresponding to FIG. 9 with a predetermined grayscale threshold value, in accordance with an embodiment of the present invention.

FIG. 11 is a schematic diagram showing an operation of removing a local mean for the gray value corresponding to FIG. 9 according to an embodiment of the present invention, and how to remove the local portion has been described in detail above. Mean, so I won't go into details here.

12 is a diagram illustrating a third comparison result obtained by comparing the luminance difference value after the partial mean subtraction operation in FIG. 11 with a predetermined luminance difference threshold value, in which the clearing can be clearly performed, according to an embodiment of the present invention. See the position of the eyeglass frame.

13 illustrates a schematic diagram of a first intersection of the first comparison result in FIG. 8 and the second comparison result in FIG. 10, wherein the second comparison result is as shown in FIG. 9 according to an embodiment of the present invention. The image component in the L channel is obtained by filtering the grayscale value corresponding to the uneven illumination intensity component and comparing it with a predetermined grayscale threshold. In Fig. 13, except for the eyeglass frame, there are almost no other unrelated areas.

14 illustrates a second intersection of the first intersection in FIG. 13 and the third comparison result in FIG. 12 (ie, the intersection of the first comparison result, the second comparison result, and the third comparison result, according to an embodiment of the present invention). The schematic diagram of the image mainly contains the iris area and few remaining fragments of the spectacle frame.

In one embodiment, performing iris extraction according to the second intersection may be implemented as follows: respectively determining whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if present, the foreground pixel a point is divided into the adjacent target pixel in the same target area to obtain a plurality of target areas; respectively calculating a number of foreground pixel points owned by the plurality of target areas; determining the most-numbered target area as an iris area; Iris extraction is performed in the iris region.

FIG. 15 illustrates a schematic diagram of a plurality of target regions obtained by dividing respective foreground pixel points into different target regions, respectively, according to an embodiment of the present invention. Based on FIG. 15, the determining step may be implemented as follows: whether each foreground pixel in the binary image is determined one above the other, below, below, left, right, upper left, upper right, lower left, and/or lower right. There is an adjacent foreground pixel, if present, dividing the foreground pixel into the same target area in the same target area, for example, using a uniform connectivity flag for each target area (eg, as an example in FIG. 15 The

numbers

1, 2, 3) are used; if not present, the foreground pixel is the separated foreground pixel and is typically the residual background area in the non-iris region.

FIG. 16 illustrates the owned foreground pixels of FIG. 15 in accordance with an embodiment of the present invention. A schematic diagram of the iris region having the largest number of dots, wherein the target region located at the intermediate portion has the largest number of foreground pixel points, and thus the target region is determined to be the iris region.

In one embodiment, holes in the iris region, non-adjacent boundaries can also be eliminated to obtain a complete iris region; and iris extraction is performed in the intact iris region. Specifically, the holes in the iris area and the non-adjacent boundaries are eliminated according to the following formula:

Represents a two-dimensional integer grid.

The following is a detailed description of how these formulas eliminate holes and non-contiguous boundaries in the iris region.

Assuming that A represents the iris region and B represents a set of structural elements with a radius R (for example, a circle with a radius of 3), the above formula can be visually understood as sliding the center of B along the edge point of A to obtain an overlapping region; The region performs a union operation with B to obtain an iris region (also referred to as an expanded iris region) whose edges are enlarged, holes, and non-adjacent boundaries are eliminated. Then, the center of B is slid along the edge point of the expanded iris region to obtain another overlapping region; the other overlapping region is intersected with B to obtain the above-mentioned complete iris region (also referred to as corrosion). The iris area), the size of the iris area remains the same, but the holes, non-adjacent boundaries that may appear inside the iris area are eliminated.

Figure 17 illustrates a schematic diagram of a complete iris region resulting from the elimination of holes in the iris region, non-adjacent boundaries, in accordance with an embodiment of the present invention, whereby it can be seen that a complete iris region without any interference, holes, has been obtained.

Those skilled in the art will appreciate that the above-described operations involving the removal of the remaining segments of the eyeglass frame by image morphology post-processing are merely exemplary and not limiting, and any other embodiments capable of achieving the removal of the remaining segments of the eyeglass frame should be incorporated into the present invention. Protection Wai.

It should be appreciated by those skilled in the art that the above-described operations relating to determining the position of the eyeglass frame are merely exemplary and not limiting, and any other embodiment capable of realizing the position of the eyeglass frame should be included in the scope of protection of the present invention.

18 illustrates a block diagram of a structure of an iris extraction apparatus under non-uniform illumination conditions, including a filtering device 1802, a computing device 1804, and an iris extraction device 1806, in accordance with an embodiment of the present invention. Wherein the filtering device 1802 is configured to filter out the uneven illumination intensity component from the image containing the iris, the computing device 1804 is configured to calculate the color difference in the filtered image of the filtering device 1802, and the iris extraction device 1806 is configured to use the computing device. The iris was extracted by the color difference calculated in 1804.

In one embodiment, the filtering device 1802 includes a conversion unit for converting a color mode of the image such that the uneven illumination intensity component and the color component in the converted image are separated.

In one embodiment, the conversion unit converts the image from the RGB color mode to the Lab color mode to obtain respective image components in the L channel, the a channel, and the b channel, wherein the image component in the L channel includes the unevenness The illumination intensity component, the image component in the a channel and the image component in the b channel include the color component. In one embodiment, computing device 1804 includes a first computing unit for calculating a color difference using image components in the a channel and image components in the b channel.

In one embodiment, the computing unit calculates the color difference using the image component in the a channel and the image component in the b channel according to the following formula:

Where ΔE _ab represents the chromatic aberration, Δ(a) represents the difference between two values in the a channel, and Δ(b) represents the difference between the two values in the b channel.

In one embodiment, the iris extraction device 1806 includes: a first comparison unit configured to compare the color difference with a predetermined color difference threshold to obtain a first comparison result; and an iris extraction unit configured to perform the first comparison result according to the first comparison result Iris extraction.

In one embodiment, the filtering device 1802 further includes a filtering unit for filtering out the uneven light intensity component from the image components in the L channel.

In one embodiment, the filtering unit includes: a first computing module for connecting the L The image component in the track performs a logarithm operation; the second operation module is configured to perform fast Fourier transform on the log transformed image component; and the third operation module is configured to filter the fast Fourier transform by high-pass filtering a low frequency portion of the image component; a fourth operation module for performing an inverse transform of the high-pass filtered image component on the fast Fourier transform; and a fifth operation module for performing an exponential operation on the inverse transformed image component.

In an embodiment, the iris extraction device 1806 further includes: a second calculation unit, configured to calculate a gray value corresponding to the image component in the L channel after filtering the uneven illumination intensity component; and a second comparison unit, configured to The gray value is compared with a predetermined gray threshold to obtain a second comparison result; a third calculating unit is configured to calculate a first intersection of the first comparison result and the second comparison result; and an iris extraction unit is configured to use the first Intersection for iris extraction. As has been explained in detail above, the execution of the iris extraction device 1806 can be performed here without wearing glasses.

In one embodiment, the iris extraction device 1806 further includes: a fourth calculation unit configured to calculate a brightness of any one of the image components in the L channel after filtering the uneven illumination intensity component minus a predetermined number of the points around the point a difference between the average values of the brightness of the points; a third comparing unit for comparing the difference with a predetermined brightness difference threshold to obtain a third comparison result; and a fourth calculating unit for calculating the first intersection and the first a second intersection of the three comparison results; and an iris extraction unit for performing iris extraction based on the second intersection.

In one embodiment, the fourth calculation unit calculates the difference according to the following formula:

Where (x, y) represents the coordinates of any point in the image component in the L channel after filtering out the uneven illumination intensity component, and f(x, y) represents the luminance value of the point of the coordinate (x, y),

Represents a non-negative integer set and n represents a non-negative integer.

In an embodiment, the iris extraction device 1806 further includes: a determining unit, configured to respectively determine whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if present, the foreground Pixel points are divided into the same target area in the same target area to obtain a plurality of target areas; and a fifth calculating unit is configured to separately calculate the number of foreground pixel points owned by the plurality of target areas; The element is used to determine the target area of the largest number as the iris area; and the iris extraction unit performs iris extraction in the iris area.

In one embodiment, the iris extraction device 1806 further includes: an elimination unit for eliminating holes in the iris region, non-adjacent boundaries, to obtain a complete iris region; and an iris extraction unit for iris extraction in the intact iris region .

Represents a two-dimensional integer grid.

In summary, according to the above embodiment of the present invention, by filtering out different uneven illumination intensity components, the influence of the uneven illumination condition on the iris extraction can be avoided; the local mean calculation can be performed by removing the image component in the L channel. The position of the spectacle frame is obtained, thereby eliminating the effect of the spectacle frame on iris extraction.

Although the invention has been described with reference to a particular embodiment thereof, it is understood that the invention is not limited to the specific embodiments disclosed. The invention is intended to cover various modifications and equivalents The scope of the following claims is to be accorded

Claims

An iris extraction method under uneven illumination conditions, comprising:

Filtering out uneven illumination intensity components from images containing irises;

Calculating the color difference in the filtered image;

Iris extraction is performed according to the color difference.
The method of claim 1 wherein said filtering out the uneven illumination intensity component from the image comprising the iris comprises:

The color mode of the image is converted such that the uneven illumination intensity component and the color component in the converted image are separated.
The method of claim 2, wherein converting the image from an RGB color mode to a Lab color mode results in respective image components in the L channel, the a channel, and the b channel such that the converted image is in the image The uneven illumination intensity component and the color component are separated, wherein the image component in the L channel includes the uneven illumination intensity component, and the image component in the a channel and the image component in the b channel include The color component.
The method of claim 3 wherein said calculating the color difference in said filtered image comprises:

The color difference is calculated using image components in the a channel and image components in the b channel.
The method of claim 4, wherein the color difference is calculated using image components in the a channel and image components in the b channel according to the following formula:

Where ΔE ab represents the chromatic aberration, Δ(a) represents the difference between the two values in the a channel, and Δ(b) represents the difference between the two values in the b channel.
The method according to claim 4 or 5, wherein said performing iris extraction based on said color difference comprises:

Comparing the color difference with a predetermined color difference threshold to obtain a first comparison result;

Iris extraction is performed based on the first comparison result.
The method of claim 3 wherein said filtering out the uneven illumination intensity component from the image comprising the iris further comprises:

The uneven illumination intensity component is filtered from image components in the L channel.
The method of claim 7 wherein filtering the non-uniform illumination intensity component from image components in the L channel comprises:

Performing a logarithmic operation on image components in the L channel;

Performing a fast Fourier transform on the log transformed image component;

Filtering the low frequency portion of the image component after the fast Fourier transform by high pass filtering;

Performing an inverse transform of the high-pass filtered image component on the fast Fourier transform;

The inversely transformed image component is subjected to an exponential operation.
The method of claim 7, wherein said performing iris extraction based on said color difference further comprises:

Comparing the color difference with a predetermined color difference threshold to obtain a first comparison result;

Calculating a gray value corresponding to an image component in the L channel after filtering the uneven light intensity component;

Comparing the gray value with a predetermined gray threshold to obtain a second comparison result;

Calculating a first intersection of the first comparison result and the second comparison result;

Iris extraction is performed according to the first intersection.
The method of claim 9, wherein said performing iris extraction based on said color difference further comprises:

Calculating a difference between an average value of luminances of the image components in the L channel after filtering the uneven illumination intensity component minus an average of luminances of a predetermined number of points around the point;

Comparing the difference with a predetermined brightness difference threshold to obtain a third comparison result;

Calculating a second intersection of the first intersection and the third comparison result;

Iris extraction is performed according to the second intersection.
The method of claim 10 wherein said difference is calculated according to the following formula:

Where (x, y) represents the coordinates of any point in the image component in the L channel after filtering out the uneven illumination intensity component, and f(x, y) represents the point of the coordinate (x, y) Brightness value,
Represents a non-negative integer set and n represents a non-negative integer.
The method according to claim 10 or 11, wherein extracting the iris according to the second intersection comprises:

Determining, respectively, whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if present, dividing the foreground pixel and the adjacent foreground pixel into the same target area Thereby obtaining a plurality of target areas;

Calculating, respectively, the number of foreground pixel points owned by the plurality of target regions;

Determining the target area of the largest number as the iris area;

Iris extraction is performed in the iris region.
The method of claim 12, wherein performing iris extraction in the iris region comprises:

Eliminating holes and non-adjacent boundaries in the iris region to obtain a complete iris region;

Iris extraction is performed in the intact iris region.
The method of claim 13 wherein the holes in the iris region, non-adjacent boundaries are eliminated according to the following formula:

Where A represents the iris region, α represents any point in A, B represents a collection of structural elements, and b represents any point in B.
Represents a two-dimensional integer grid.
The method of claim 14 wherein the iris is further eliminated according to the following formula Holes in the area, non-adjacent boundaries:

Where A represents the iris region, α represents any point in A, and B represents a collection of structural elements.
Represents a two-dimensional integer grid.
An iris extraction device under uneven illumination conditions, comprising:

a filtering device for filtering out uneven light intensity components from the image containing the iris;

a computing device for calculating a color difference in the filtered image;

An iris extraction device for performing iris extraction according to the chromatic aberration.
The apparatus of claim 16 wherein said filtering means comprises:

And a converting unit, configured to convert a color mode of the image, so that the uneven light intensity component and the color component in the converted image are separated.
The apparatus according to claim 17, wherein said converting unit converts said image from an RGB color mode to a Lab color mode, and obtains corresponding image components in the L channel, the a channel, and the b channel, so that the converted image The uneven illumination intensity component and the color component in the image are separated, wherein the image component in the L channel includes the uneven illumination intensity component, an image component in the a channel, and the b channel The image component includes the color component.
The device of claim 18 wherein said computing device comprises:

a first calculating unit, configured to calculate the color difference by using an image component in the a channel and an image component in the b channel.
The apparatus according to claim 19, wherein said first calculating unit calculates said color difference using an image component in said a channel and an image component in said b channel according to the following formula:

Where ΔE ab represents the chromatic aberration, Δ(a) represents the difference between the two values in the a channel, and Δ(b) represents the difference between the two values in the b channel.
The apparatus according to claim 19 or 20, wherein said iris extraction means comprises:

a first comparing unit, configured to compare the color difference with a predetermined color difference threshold To the first comparison result;

An iris extraction unit configured to perform iris extraction according to the first comparison result.
The apparatus according to claim 18, wherein said filtering means further comprises:

And filtering a unit for filtering the uneven light intensity component from image components in the L channel.
The apparatus of claim 22 wherein said filtering unit comprises:

a first operation module, configured to perform logarithm operation on image components in the L channel;

a second operation module, configured to perform fast Fourier transform on the log transformed image component;

a third operation module, configured to filter, by high-pass filtering, a low-frequency portion of the image component after fast Fourier transform;

a fourth operation module, configured to perform inverse transform of the high-pass filtered image component by fast Fourier transform;

And a fifth operation module, configured to perform an exponential operation on the inverse transformed image component.
The apparatus according to claim 22, wherein said iris extraction means comprises:

a first comparing unit, configured to compare the color difference with a predetermined color difference threshold to obtain a first comparison result;

a second calculating unit, configured to calculate a gray value corresponding to the image component in the L channel after filtering the uneven light intensity component;

a second comparing unit, configured to compare the gray value with a predetermined gray threshold to obtain a second comparison result;

a third calculating unit, configured to calculate a first intersection of the first comparison result and the second comparison result;

The iris extraction unit is configured to perform iris extraction according to the first intersection.
The apparatus according to claim 24, wherein said iris extracting means further comprises:

a fourth calculating unit, configured to calculate a difference between an average value of brightness of the image component in the L channel after filtering the uneven light intensity component minus a brightness of a predetermined number of points around the point Value

a third comparing unit, configured to compare the difference with a predetermined brightness difference threshold to obtain a third comparison result;

a fourth calculating unit, configured to calculate a second intersection of the first intersection and the third comparison result;

The iris extraction unit is configured to perform iris extraction according to the second intersection.
The apparatus according to claim 25, wherein said fourth calculating unit calculates said difference value according to the following formula:

Where (x, y) represents the coordinates of any point in the image component in the L channel after filtering out the uneven illumination intensity component, and f(x, y) represents the point of the coordinate (x, y) Brightness value,
Represents a non-negative integer set and n represents a non-negative integer.
The apparatus according to claim 25 or 26, wherein the iris extraction device further comprises:

a determining unit, configured to determine, respectively, whether each foreground pixel in the binary image corresponding to the second intersection has an adjacent foreground pixel, and if present, the foreground pixel and the adjacent foreground pixel Divided into the same target area to obtain multiple target areas;

a fifth calculating unit, configured to separately calculate a number of foreground pixel points owned by the plurality of target areas;

a determining unit, configured to determine the target area of the largest number as an iris area;

The iris extraction unit performs iris extraction in the iris region.
The apparatus according to claim 27, wherein the iris extraction device further comprises:

a eliminating unit for eliminating holes in the iris region, non-adjacent boundaries, and obtaining a complete iris region;

The iris extraction unit performs iris extraction in the intact iris region.
The apparatus according to claim 28, wherein said eliminating unit eliminates holes, non-adjacent boundaries in said iris region according to the following formula:

Where A represents the iris region, α represents any point in A, B represents a collection of structural elements, and b represents any point in B.
Represents a two-dimensional integer grid.
The apparatus according to claim 29, wherein said eliminating unit further eliminates holes, non-adjacent boundaries in the iris region according to the following formula:

Where A represents the iris region, α represents any point in A, and B represents a collection of structural elements.
Represents a two-dimensional integer grid.