CN107437064B - Iris living body detection method based on spectral analysis - Google Patents
Iris living body detection method based on spectral analysis Download PDFInfo
- Publication number
- CN107437064B CN107437064B CN201710540599.1A CN201710540599A CN107437064B CN 107437064 B CN107437064 B CN 107437064B CN 201710540599 A CN201710540599 A CN 201710540599A CN 107437064 B CN107437064 B CN 107437064B
- Authority
- CN
- China
- Prior art keywords
- iris
- living body
- living
- detection
- eye
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/40—Spoof detection, e.g. liveness detection
- G06V40/45—Detection of the body part being alive
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/18—Eye characteristics, e.g. of the iris
- G06V40/193—Preprocessing; Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Multimedia (AREA)
- Human Computer Interaction (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Ophthalmology & Optometry (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
Abstract
The invention discloses an iris living body detection method based on spectral analysis. The invention trains the frequency spectrum distribution characteristics of the living iris different from the printed iris by using the frequency spectrum distribution characteristic difference of the living iris and the printed iris, realizes the classification of the living iris and the printed iris, effectively prevents the false body attack phenomenon of imitating the living iris in the identity authentication process, and is suitable for maintaining the information security of users. The invention sets various iris living body detection judgment mechanisms which can be configured according to actual requirements.
Description
Technical Field
The invention relates to the technical fields of digital image processing, pattern recognition, statistical learning and the like, in particular to an iris living body detection method based on spectral analysis.
Background
The iris identification identifies and authenticates the identity of a person by analyzing the texture difference between different irises, has the advantages of high uniqueness, strong stability, non-invasion and the like, and is successfully applied to the identity identification of occasions such as airports, customs, banks and the like. However, with the increasing popularity, the iris recognition system is also confronted with threats and attacks from various artificial counterfeiting technologies. If the iris recognition system is unable to accurately detect and alert counterfeit iris data, potential losses will result to the authorized user.
Therefore, how to quickly and effectively realize the living iris detection in the iris recognition system still is a problem to be solved urgently.
Disclosure of Invention
The invention aims to provide a quick and effective iris in-vivo detection method aiming at the problems in the prior art.
An iris living body detection method based on spectral analysis comprises a living body iris frequency spectrum distribution characteristic training process S1 and an iris living body detection process S2.
The living body iris spectrum distribution characteristic training process S1 comprises the following sub-steps: constructing a living iris database as a training library, preprocessing images in the training library, and cutting out an eye region; extracting eye region features by using a frequency spectrum transformation method to obtain an eye spectrogram; obtaining a binary difference image of the eye spectrogram by adopting an image processing method; carrying out region analysis to obtain an eye frequency spectrum analysis chart; overlapping eye frequency spectrum analysis images of all images in a training library, and counting a living iris frequency spectrum energy distribution map; and performing multi-stage decomposition on the living body iris frequency spectrum energy distribution map to obtain a multi-stage living body iris frequency spectrum energy distribution map.
The iris liveness detection process S2 includes the following sub-steps: preprocessing the test image and cutting out an eye image; extracting eye region features by using a frequency spectrum transformation method to obtain an eye spectrogram; obtaining a binary difference image of the eye spectrogram by adopting an image processing method; selecting a judgment mechanism to carry out living body detection according to the multi-stage living body iris frequency spectrum energy distribution map obtained by training; and outputting the result of the living body detection.
Preferably, the living iris spectrum distribution characteristic training process S1 includes the following specific steps:
s11: preprocessing images in a training library, positioning the positions of eyes by adopting a human eye detection method, and cutting out eye images;
s12: extracting the spectral characteristics of the eye region by using a spectral transformation method of fast Fourier transform to obtain an eye spectrogram;
s13: carrying out image processing on the eye spectrogram, wherein the image processing comprises filtering, difference value processing and binarization processing to obtain a binarized difference image of the eye spectrogram;
s14: performing region analysis on the binarized difference image by adopting a connected domain analysis method, and reserving a region with a larger connected domain area to obtain a spectrum analysis chart;
s15: overlapping eye frequency spectrum analysis images of all images in a training library, and counting a living iris frequency spectrum energy distribution map;
s16: and performing multi-stage decomposition on the living body iris frequency spectrum energy distribution map to obtain a strong and weak living body iris frequency spectrum energy distribution map.
The multi-stage decomposition method of step S16 includes: setting thresholds a and b, wherein a is larger than b, marking the region with the value larger than b of the living iris spectrum energy distribution diagram as 1, otherwise marking the region as 0, and marking the marked image as a weak living iris spectrum energy distribution diagram; and marking the area with the value of the living iris spectrum energy distribution map larger than a as 1, otherwise marking the area as 0, and marking the marked image as a strong living iris spectrum energy distribution map.
Preferably, the iris biopsy procedure S2 includes the following steps:
s21: preprocessing a test image, positioning the position of eyes by adopting a human eye detection method, and cutting out an eye image;
s22: extracting the spectral characteristics of the eye region by using a spectral transformation method of fast Fourier transform to obtain an eye spectrogram;
s23: carrying out image processing on the eye spectrogram, wherein the image processing comprises filtering, difference value processing and binarization processing to obtain a binarized difference image of the eye spectrogram;
s24: selecting a judgment mechanism to carry out living body detection according to the multi-stage living body iris frequency spectrum energy distribution map obtained by training;
s25: and outputting the result of the living body detection.
Preferably, the decision mechanism of step S24 in the iris liveness detection process S2 is set as the following steps:
s2411: carrying out weak living body iris detection, carrying out filtering processing on the binarized difference image obtained in the step S23 by utilizing a weak living body iris frequency spectrum energy distribution map obtained by training, carrying out connected domain analysis, and reserving a region with a larger connected domain area;
s2412: judging whether the iris is a weak living body iris, if so, continuing to detect the strong living body iris, and entering a step S2413, otherwise, judging that a prosthesis is detected, sending an alarm, and terminating the iris living body detection process;
s2413: performing strong living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2411 by using a strong living body iris frequency spectrum energy distribution map obtained by training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2414: and judging whether the iris is a strong living iris, if so, judging the iris to be the living iris, outputting a detection result, terminating the iris living detection process, and otherwise, returning to the step S21 to carry out iris living detection on the next test image.
Preferably, the decision mechanism of step S24 in the iris liveness detection process S2 may be further configured as the following steps:
s2421: strong living body iris detection is carried out, the binaryzation difference image obtained in the step S23 is subjected to filtering processing by utilizing a strong living body iris frequency spectrum energy distribution map obtained through training, connected domain analysis is carried out, and a region with a large connected domain area is reserved;
s2422: judging whether the iris is a strong living iris, if so, judging the iris to be a living iris, outputting a detection result, terminating the living iris detection process, otherwise, continuing to perform weak living iris detection, and entering the step S2423;
s2423: performing weak living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2421 by using the frequency spectrum energy distribution map of the weak living body iris obtained through training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2424: and judging whether the iris is a weak living body, if so, performing iris living body detection on the next frame of image, returning to the step S21 to perform iris living body detection on the next test image, otherwise, judging that a false body is detected, giving an alarm, and stopping the iris living body detection process.
Preferably, the step S2422 of the decision mechanism is to count the number of times of continuously judging as not being a living iris after judging as not being a living iris, and when the number of times exceeds a predetermined threshold, directly judging as a false body, giving an alarm, and terminating the iris living body detection process.
Preferably, the filtering processes of steps S13 and S23 include two filtering processes, in which the image is filtered once by a filter with a radius of 1 pixel, and the image is filtered twice by a filter with a radius of 24 pixels.
Preferably, the difference processing method in step S13 and step S23 is to perform a difference on the primary filtered image and the secondary filtered image pixel by pixel, take an absolute value, and multiply the absolute value by a scaling factor.
The invention has the beneficial effects that: the invention trains the frequency spectrum distribution characteristics of the living iris different from the printed iris by using the frequency spectrum distribution characteristic difference of the living iris and the printed iris, realizes the classification of the living iris and the printed iris, detects the false body attack phenomenon imitating the living iris in the identity authentication process, and is suitable for maintaining the information security of users. The invention sets various iris living body detection judgment mechanisms which can be configured according to actual requirements.
It is to be understood that both the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention, as claimed.
Drawings
Further objects, features and advantages of the present invention will become apparent from the following description of embodiments of the invention, with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of an iris biopsy method based on spectral analysis according to the present invention;
FIG. 2a shows a living iris detection method based on spectral analysis according to a first embodiment of the present invention;
FIG. 2b shows a living iris detection method based on spectral analysis according to a second embodiment of the present invention;
FIG. 2c shows a third embodiment iris biopsy method based on spectral analysis according to the present invention;
FIG. 3a schematically illustrates a live iris detection of an iris in accordance with the present invention;
FIG. 3b schematically illustrates printing an iris in accordance with the present invention;
FIG. 3c is a graph of the spectrum of a live iris 3a shown for live iris detection in accordance with the present invention;
fig. 3d is a spectrum diagram corresponding to a printed iris 3b shown for iris liveness detection in accordance with the present invention.
Detailed Description
The objects and functions of the present invention and methods for accomplishing the same will be apparent by reference to the exemplary embodiments. However, the present invention is not limited to the exemplary embodiments disclosed below; it can be implemented in different forms. The nature of the description is merely to assist those skilled in the relevant art in a comprehensive understanding of the specific details of the invention.
Hereinafter, embodiments of the present invention will be described with reference to the accompanying drawings. In the drawings, the same reference numerals denote the same or similar parts, or the same or similar steps.
Fig. 1 shows basic steps of an iris living body detection method based on spectral analysis, and as shown in fig. 1, the method includes a living body iris spectral distribution characteristic training process S1 and an iris living body detection process S2 based on spectral analysis, and the living body iris spectral distribution characteristic training process S1 includes the following specific steps:
step S11 is to pre-process the images in the training library, locate the eye position by using the human eye detection method, and cut out the eye area.
Step S12 is to extract the spectral features of the eye region by using a fast fourier transform method to obtain an eye spectrogram, further perform image enhancement processing, set an image enhancement threshold T1, and set the pixel value of a point smaller than a preset threshold in the eye spectrogram as T1.
Step S13 is to perform image processing on the eye spectrogram, including filtering, difference processing, and binarization processing, to obtain a binarized difference image of the eye spectrogram. According to the symmetry of the spectrogram, only the upper half region of the spectrogram can be subjected to filtering processing. The filtering processing method comprises mean filtering, median filtering and the like, has two filtering processes, and adopts a filter with the radius of 1 pixel to carry out primary filtering on the image and adopts a filter with the radius of 24 pixels to carry out secondary filtering on the image. The difference processing method is to carry out difference on the primary filtering image and the secondary filtering image pixel by pixel, take the absolute value and then multiply the absolute value by a scaling coefficient.
And step S14, performing region analysis on the binarized difference image by adopting a connected domain analysis method, and reserving a region with a larger area of a connected domain to obtain a spectrum analysis chart.
Step S15 is to superimpose the eye spectrum analysis maps of all images in the training library, and count the living iris spectrum energy distribution map. The living body iris spectrum energy distribution graph represents the frequency of the living body iris spectrum energy, and the larger the value of a certain position in the living body iris spectrum energy distribution graph is, the higher the probability of the living body iris spectrum energy at the position is. That is, the spectral energy of the living iris is concentrated in the region where the value of the spectral energy distribution of the living iris is high; the spectral energy of the printed iris is not within the region of higher value of the spectral energy profile of the living iris and there are multiple regions of aggregated energy in the spectral energy profile of the printed iris.
And step S16, performing multi-stage decomposition on the living body iris frequency spectrum energy distribution map, setting the living body iris frequency spectrum energy distribution map into two stages, and obtaining a strong living body iris frequency spectrum energy distribution map and a weak living body iris frequency spectrum energy distribution map. Setting threshold values a and b, wherein a is larger than b, marking a region with the value of the living body iris spectrum energy distribution diagram larger than b as 1, otherwise marking the region as 0, and marking the marked image as a weak living body iris spectrum energy distribution diagram; and marking the area with the value of the living iris spectrum energy distribution map larger than a as 1, otherwise marking the area as 0, and marking the marked image as a strong living iris spectrum energy distribution map.
The region with a value of 1 in the weak live iris spectral energy profile indicates the location where the spectral energy of the live iris and the partially printed iris occurs, and the region with a value of 0 in the weak live iris spectral energy profile indicates the location where the spectral energy of the printed iris occurs.
The region with the value of 1 in the strong living iris spectral energy distribution map indicates the position where the spectral energy of the living iris appears, and the region with the value of 0 in the strong living iris spectral energy distribution map indicates the position where the spectral energy of the printed iris and part of the living iris appears.
The iris living body detection process S2 based on the spectrum analysis comprises the following specific steps:
step S21 is to preprocess the test image, locate the eye position by using the eye detection method, and cut out the eye image. If no eyes are detected, the next test image may be pre-processed.
Step S22 is to extract the spectral features of the eye region by using a fast fourier transform method to obtain an eye spectrogram, further perform image enhancement processing, set an image enhancement threshold T1, and set the pixel value of a point smaller than a preset threshold in the eye spectrogram as T1.
Step S23 is to perform image processing on the eye spectrogram, including filtering, difference processing, and binarization processing, to obtain a binarized difference image of the eye spectrogram. According to the symmetry of the spectrogram, only the upper half region of the spectrogram can be subjected to filtering processing. The filtering processing method comprises mean filtering, median filtering and the like, has two filtering processes, and adopts a filter with the radius of 1 pixel to carry out primary filtering on the image and adopts a filter with the radius of 24 pixels to carry out secondary filtering on the image. The difference processing method is to carry out difference on the primary filtering image and the secondary filtering image pixel by pixel, take the absolute value and then multiply the absolute value by a scaling coefficient.
Step S24 is to select a decision mechanism for live body detection according to the multi-level live body iris spectrum energy distribution map obtained by training, where the decision mechanism includes 3 types.
Step S25 is outputting the result of the live body detection, and in the process of iris recognition identity authentication, if the output is a live iris, the current user can continue iris registration and iris recognition, and if the output is a false body, the current user is not authorized to perform iris registration and iris recognition.
Example 1
The 1 st judgment mechanism is used for detecting whether the iris is a prosthesis or not through weak living body iris detection of a first-level detection strategy aiming at an application scene with frequent prosthesis attack, so that the condition of a large number of prosthesis attacks can be eliminated, then strong living body iris detection of a second-level detection strategy is carried out on a small number of irises judged to be weak living bodies, and finally whether the iris is a living body iris or not is detected. The detection strategy has the advantages that the condition of the prosthesis attack can be detected only through the first-level detection strategy, the prosthesis attack behavior can be rapidly and efficiently sent out to give an alarm in the identity authentication process, and the security of the identity authentication is ensured. The specific steps are as in S2411 to S2414 of fig. 2 a:
s2411: carrying out weak living body iris detection, carrying out filtering processing on the binarized difference image obtained in the step S23 by utilizing a weak living body iris frequency spectrum energy distribution map obtained by training, carrying out connected domain analysis, and reserving a region with a larger connected domain area;
s2412: judging whether the iris is a weak living body iris, if so, continuing to detect the strong living body iris, and entering a step S2413, otherwise, judging that a prosthesis is detected, sending an alarm, and terminating the iris living body detection process;
s2413: performing strong living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2411 by using a strong living body iris frequency spectrum energy distribution map obtained by training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2414: and judging whether the iris is a strong living iris, if so, judging the iris to be the living iris, outputting a detection result, terminating the iris living detection process, and otherwise, returning to the step S21 to carry out iris living detection on the next test image.
Example 2
The 2 nd judgment mechanism is used for detecting whether the iris is a living iris or not through strong living iris detection of a first-stage detection strategy aiming at an application scene in which false body attack does not occur frequently, and detecting whether the iris is a false body or not through weak living iris detection of a second-stage detection strategy after a small part of images which cannot be judged as the living iris are remained. The detection strategy has the advantages that the living body iris can be detected only through the first-level detection strategy, the corresponding authority is given to the user for identity authentication, and the calculation complexity of the algorithm is reduced in the process of identity authentication of the living body iris user. The specific steps are as in S2421 to S2424 of fig. 2 b:
s2421: strong living body iris detection is carried out, the binaryzation difference image obtained in the step S23 is subjected to filtering processing by utilizing a strong living body iris frequency spectrum energy distribution map obtained through training, connected domain analysis is carried out, and a region with a large connected domain area is reserved;
s2422: judging whether the iris is a strong living iris, if so, judging the iris to be a living iris, outputting a detection result, terminating the living iris detection process, otherwise, continuing to perform weak living iris detection, and entering the step S2423;
s2423: performing weak living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2421 by using the frequency spectrum energy distribution map of the weak living body iris obtained through training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2424: and judging whether the iris is a weak living body, if so, performing iris living body detection on the next frame of image, returning to the step S21 to perform iris living body detection on the next test image, otherwise, judging that a false body is detected, giving an alarm, and stopping the iris living body detection process.
Example 3
Compared with the decision mechanism of the 2 nd decision mechanism, the decision mechanism of the 3 rd decision mechanism is more rigorous to the judgment of the living iris, namely after the strong living iris detection of the first-level detection strategy, if the times of continuously judging that the iris is not the strong living iris exceeds the preset threshold value, the iris is directly judged to be printed, alarm information is sent, and the next frame image of the current user is not subjected to the living detection any more. The specific steps are as in S2431-S2436 of FIG. 2 c:
s2431: strong living body iris detection is carried out, the binaryzation difference image obtained in the step S23 is subjected to filtering processing by utilizing a strong living body iris frequency spectrum energy distribution map obtained through training, connected domain analysis is carried out, and a region with a large connected domain area is reserved;
s2432: judging whether the iris is a strong living iris, if so, judging the iris to be a living iris, outputting a detection result, and stopping the iris living detection process, otherwise, entering the step S2433;
s2433: accumulating the times of continuously judging whether the iris is a living iris;
s2434: judging whether the number of times of continuously judging whether the iris of the living body exceeds a threshold value, if so, judging that the false body is detected, giving an alarm, terminating the iris living body detection process, otherwise, continuing to detect the iris of the weak living body, and entering the step S2435;
s2435: carrying out weak living body iris detection, carrying out filtering processing on the image subjected to filtering processing in the step S2431 by utilizing a weak living body iris frequency spectrum energy distribution map obtained by training, carrying out connected domain analysis, and reserving a region with a larger connected domain area;
s2436: and judging whether the iris is a weak living body, if so, performing iris living body detection on the next frame of image, returning to the step S21 to perform iris living body detection on the next test image, otherwise, judging that a false body is detected, giving an alarm, and stopping the iris living body detection process.
Fig. 3a to 3d are spectrum diagrams of live irises and printed irises, wherein fig. 3a is a live iris, fig. 3b is a printed iris from the live iris shown in fig. 3a, and fig. 3c and 3d are fast fourier transform spectrum diagrams of fig. 3a and 3b, respectively, and it can be seen that there is a great difference in the spectrum distribution of the live iris and the printed iris: the high frequency components in the spectrogram of the living iris are distributed more intensively, compared with the frequency spectrum of the living iris, the printed iris has more medium and high frequency components, and four bright spots regularly distributed around the frequency spectrum in the graph of fig. 3d are the frequency characteristics generated in the printing process. Therefore, the living iris and the printed iris can be distinguished according to the frequency spectrum distribution difference, the position where the living iris frequency spectrum appears is filtered, then judgment is carried out according to the number of the residual energy areas, a threshold value is set, if the number of the residual energy areas is higher than the threshold value, the iris is judged to be the printed iris, and if not, the iris is judged to be the living iris.
Other embodiments of the invention will be apparent to those skilled in the art from consideration of the specification and practice of the invention disclosed herein. It is intended that the specification and examples be considered as exemplary only, with a true scope and spirit of the invention being indicated by the following claims.
Claims (6)
1. An iris living body detection method based on spectral analysis comprises a living body iris frequency spectrum distribution characteristic training process S1 and an iris living body detection process S2;
the living body iris spectrum distribution characteristic training process S1 comprises the following substeps: constructing a living iris database as a training library, preprocessing images in the training library, and cutting out an eye region; extracting eye region features by using a frequency spectrum transformation method to obtain an eye spectrogram; obtaining a binary difference image of the eye spectrogram by adopting an image processing method; carrying out region analysis to obtain an eye frequency spectrum analysis chart; overlapping eye frequency spectrum analysis images of all images in a training library, and counting a living iris frequency spectrum energy distribution map; performing multi-stage decomposition on the living body iris frequency spectrum energy distribution map to obtain a multi-stage living body iris frequency spectrum energy distribution map;
the iris liveness detection process S2 includes the following sub-steps:
s21: preprocessing a test image, positioning the position of eyes by adopting a human eye detection method, and cutting out an eye image;
s22: extracting the spectral characteristics of the eye region by using a spectral transformation method of fast Fourier transform to obtain an eye spectrogram;
s23: carrying out image processing on the eye spectrogram, wherein the image processing comprises filtering, difference value processing and binarization processing to obtain a binarized difference image of the eye spectrogram;
s24: selecting a judgment mechanism to carry out living body detection according to the multi-stage living body iris frequency spectrum energy distribution map obtained by training;
s25: outputting a living body detection result;
wherein the decision mechanism is arranged as follows:
s2411: carrying out weak living body iris detection, carrying out filtering processing on the binarized difference image obtained in the step S23 in the iris living body detection process S2 by utilizing a weak living body iris frequency spectrum energy distribution map obtained by training, carrying out connected domain analysis, and reserving a region with a larger connected domain area;
s2412: judging whether the iris is a weak living body iris, if so, continuing to detect the strong living body iris, and entering a step S2413, otherwise, judging that a prosthesis is detected, sending an alarm, and terminating the iris living body detection process;
s2413: performing strong living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2411 by using a strong living body iris frequency spectrum energy distribution map obtained by training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2414: judging whether the iris is a strong living iris, if so, judging the iris to be a living iris, outputting a detection result, terminating the iris living detection process, and otherwise, returning to the step S21 to carry out iris living detection on the next test image;
alternatively, the decision mechanism may be further configured as the following steps:
s2421: strong living body iris detection is carried out, the binaryzation difference image obtained in the step S23 in the iris living body detection process S2 is filtered by utilizing the strong living body iris frequency spectrum energy distribution map obtained by training, connected domain analysis is carried out, and a region with a large connected domain area is reserved;
s2422: judging whether the iris is a strong living iris, if so, judging the iris to be a living iris, outputting a detection result, terminating the living iris detection process, otherwise, continuing to perform weak living iris detection, and entering the step S2423;
s2423: performing weak living body iris detection, performing filtering processing on the image subjected to filtering processing in the step S2421 by using the frequency spectrum energy distribution map of the weak living body iris obtained through training, performing connected domain analysis, and reserving a region with a larger area of the connected domain;
s2424: and judging whether the iris is a weak living body iris, if so, performing living body iris detection on the next frame image, returning to the step S21 in the living body iris detection process S2 to perform living body iris detection on the next test image, otherwise, judging that a false body is detected, giving an alarm, and terminating the living body iris detection process.
2. The iris living body detection method based on the frequency spectrum analysis as claimed in claim 1, wherein said living body iris frequency spectrum distribution characteristic training process S1 includes the following steps:
s11: preprocessing images in a training library, positioning the positions of eyes by adopting a human eye detection method, and cutting out eye images;
s12: extracting the spectral characteristics of the eye region by using a spectral transformation method of fast Fourier transform to obtain an eye spectrogram;
s13: carrying out image processing on the eye spectrogram, wherein the image processing comprises filtering, difference value processing and binarization processing to obtain a binarized difference image of the eye spectrogram;
s14: performing region analysis on the binarized difference image by adopting a connected domain analysis method, and reserving a region with a larger connected domain area to obtain a spectrum analysis chart;
s15: overlapping eye frequency spectrum analysis images of all images in a training library, and counting a living iris frequency spectrum energy distribution map;
s16: and performing multi-stage decomposition on the living body iris frequency spectrum energy distribution map to obtain a strong and weak living body iris frequency spectrum energy distribution map.
3. The iris living body detection method based on spectral analysis as claimed in claim 2, wherein the multi-stage decomposition method of step S16 in the training process S1 for iris spectrum distribution characteristics of living body is: setting thresholds a and b, wherein a is larger than b, marking the region with the value larger than b of the living iris spectrum energy distribution diagram as 1, otherwise marking the region as 0, and marking the marked image as a weak living iris spectrum energy distribution diagram; and marking the area with the value of the living iris spectrum energy distribution map larger than a as 1, otherwise marking the area as 0, and marking the marked image as a strong living iris spectrum energy distribution map.
4. The iris biopsy method based on the spectrum analysis as claimed in claim 1, wherein the iris biopsy procedure S2, the step S2422 of the decision mechanism accumulates the number of times of continuous iris determination as not being a living iris after the iris determination as not being a living iris, when the number of times exceeds a predetermined threshold, the iris biopsy procedure is terminated by directly determining as a false body and giving an alarm.
5. The iris liveness detection method based on spectral analysis as claimed in claim 1 or 2, wherein the filtering process of step S13 in the training process S1 and step S23 in the process S2 of iris liveness detection includes two filtering processes, one filtering process is performed on the image by using a filter with a radius of 1 pixel, and the other filtering process is performed on the image by using a filter with a radius of 24 pixels.
6. The iris liveness detection method based on spectral analysis as claimed in claim 1 or 2, wherein the difference processing method of step S13 in the training process S1 and step S23 in the process S2 of iris liveness detection is to make a difference pixel by pixel for the primary filtered and secondary filtered images of the iris liveness detection method based on spectral analysis, take the absolute value, and multiply by the scaling factor.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710540599.1A CN107437064B (en) | 2017-07-05 | 2017-07-05 | Iris living body detection method based on spectral analysis |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710540599.1A CN107437064B (en) | 2017-07-05 | 2017-07-05 | Iris living body detection method based on spectral analysis |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN107437064A CN107437064A (en) | 2017-12-05 |
| CN107437064B true CN107437064B (en) | 2020-11-20 |
Family
ID=60460852
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710540599.1A Active CN107437064B (en) | 2017-07-05 | 2017-07-05 | Iris living body detection method based on spectral analysis |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107437064B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109472768B (en) * | 2018-09-19 | 2022-02-25 | 上海泛洲信息科技有限公司 | Method for distinguishing object and non-object plane images by using frequency spectrum analysis |
| CN109684966A (en) * | 2018-12-17 | 2019-04-26 | 北京中科虹霸科技有限公司 | A kind of iris capturing apparatus and method for of extended field of view range |
| CN111507198B (en) * | 2020-03-25 | 2023-11-28 | 北京万里红科技有限公司 | Training method for printing iris detection model, and printing iris detection method and device |
| CN111814564A (en) * | 2020-06-09 | 2020-10-23 | 广州视源电子科技股份有限公司 | Multispectral image-based living body detection method, device, equipment and storage medium |
| CN112733791B (en) * | 2021-01-21 | 2022-09-30 | 展讯通信（上海）有限公司 | Living body detection method and device, storage medium and terminal |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1426760A (en) * | 2001-12-18 | 2003-07-02 | 中国科学院自动化研究所 | Identity discriminating method based on living body iris |
-
2017
- 2017-07-05 CN CN201710540599.1A patent/CN107437064B/en active Active
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1426760A (en) * | 2001-12-18 | 2003-07-02 | 中国科学院自动化研究所 | Identity discriminating method based on living body iris |
Non-Patent Citations (7)
| Title |
|---|
| A Fake Iris Detection Method Based on FFT and Quality Assessment;Xiaofu He;《2008 Chinese Conference on Pattern Recognition》;20081014;全文 * |
| Live face detection based on the analysis of Fourier spectra;Li;《PROCEEDINGS OF SPIE》;20040825;全文 * |
| Liveness detection for iris recognition using multispectral images;Rui Chen;《Pattern Recognition Letters》;20120901;全文 * |
| 人眼检测及瞳孔定位;余龙华;《计算机工程与应用》;20130215;全文 * |
| 虹膜图像质量评估算法的研究;姚翠莉;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120615(第6期);全文 * |
| 虹膜识别算法研究;周水子;《中国优秀硕士学位论文全文数据库 信息科技辑》;20170215;第2章 * |
| 非理想成像条件下的虹膜定位方法;吴国龙;《计算机工程与设计》;20141130;第35卷(第11期);全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN107437064A (en) | 2017-12-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN107437064B (en) | Iris living body detection method based on spectral analysis | |
| CN111738230B (en) | Face recognition method, face recognition device and electronic equipment | |
| Chen et al. | Iris recognition based on human-interpretable features | |
| Proença et al. | UBIRIS: A noisy iris image database | |
| Zhang et al. | Robust palmprint verification using 2D and 3D features | |
| Johnson et al. | Fingerprint pore characteristics for liveness detection | |
| KR101443139B1 (en) | Single image-based fake face detection | |
| CN105488486A (en) | Face recognition method and device for preventing photo attack | |
| CN109325959A (en) | A kind of method and its application of the extraction infrared image details based on Hough transform | |
| Asmuni et al. | An improved multiscale retinex algorithm for motion-blurred iris images to minimize the intra-individual variations | |
| Shen et al. | Iris matching by crypts and anti-crypts | |
| Ives et al. | Iris recognition using the ridge energy direction (RED) algorithm | |
| Haware et al. | Retina based biometric identification using SURF and ORB feature descriptors | |
| Satapathy et al. | Generalized benford’s law for fake fingerprint detection | |
| Teja | Real-time live face detection using face template matching and DCT energy analysis | |
| Hadiprakoso | Face anti-spoofing method with blinking eye and hsv texture analysis | |
| Ibrahim | Iris recognition using Haar wavelet transform | |
| Deb et al. | Vehicle license plate detection algorithm based on color space and geometrical properties | |
| CN105574481B (en) | A kind of iris image recognition methods and user equipment | |
| Shen et al. | Using crypts as iris minutiae | |
| Deshpande et al. | Iris recognition system using block based approach with DWT and DCT | |
| Aydoğdu et al. | A study on liveness analysis for palmprint recognition system | |
| Hongo et al. | Personal authentication with an iris image captured under visible-light condition | |
| Kovoor et al. | Iris biometric recognition system employing canny operator | |
| Choras | Biometric personal authentication using images of forearm vein patterns |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |