WO2009072859A2

WO2009072859A2 - System and method for authenticating image liveness

Info

Publication number: WO2009072859A2
Application number: PCT/MY2008/000170
Authority: WO
Inventors: Weng Kin Lai; Thomas Henrique Maul
Original assignee: Mimos Berhad
Priority date: 2007-12-03
Filing date: 2008-12-01
Publication date: 2009-06-11
Also published as: WO2009072859A3; MY142600A

Abstract

A method of processing image for authenticating of image liveness is identified by detecting a fake image based on the background light conditions and determining its liveness by analyzing the properties of the lighting source. The system includes image sensor for capturing image from a scene for a flicker detecting unit to detect the presence of flicker in the image to identify inactivity and the captured image is then fed to a flicker analyzing unit to determine whether the image that is being processed is actually a live image and not just a replaced static image.

Description

System and Method for Authenticating Image Liveness

Field of Invention

The present invention relates generally to a system and method for authenticating image liveness and more particularly to a system and method of detecting flickering during periods of inactivity in an indoor scene and discriminating live scene from fraudulent cases.

Background of the Invention

It is well known to provide a plurality of capture devices placed at various locations within a building or a room to provide a surveillance system for monitoring or capturing unauthorized activities within the building or room. The capture devices such as cameras are used to capture still or moving images, where each capture device is connected to a respective display screen on which the images captured by the devices can be viewed. However, intruders still attempt to enter such protected areas with the purpose of committing criminal activities.

Many video surveillance systems rely on humans to monitor the images relayed by the cameras round the clock. Due to the nature of the environment that is being continuously monitored, there may not be any significant changes that the human eye can detect, especially during periods of low human activity. During such periods of low activity, the live environment under surveillance may be replaced by a fake image, thereby giving rise to the possibility that any unauthorized activities would be missed.

For commercial properties, it is not uncommon to use fluorescent lamps to light up rooms, vaults, walkways, etc. The popularity of fluorescent lighting is on the rise as consumers are becoming more environmentally conscious and are looking for energy saving solutions. Schools and businesses find the cost savings of fluorescents to be significant.

Fluorescent lamps which operate directly from mains frequency AC will flicker at twice the mains frequency, since the power being delivered to the lamp drops to zero twice per cycle. This means the light flickers at 120 times per second (Hz) in countries which use 60-cycle-per-second (60 Hz) AC, and 100 times per , second in those which use 50 Hz.

Therefore, the present invention provides a method and system to detect and analyze the flickering of fluorescent lights in video images for the purpose of liveness-detection. The present invention also aims to exploit every possible statistical difference of flickering phenomena that holds between the live and fraudulent scenes.

The object of the present invention is to provide a method and apparatus for processing the video images to identify inactivity so as to determine whether the image that is being processed is actually a live image and not a replaced static image.

Other objects of this invention will become apparent on the reading of this entire disclosure.

Summary of the Invention

In the present invention a liveness detection system comprises an image sensor for capturing a scene, a flicker detecting unit coupled to the image sensor that detects a periodic pattern of fluctuation in an image captured by the image sensor which indicate the presence of flicker, and a flicker analyzing unit coupled to the image sensor and flicker detecting unit to perform illumination analysis on the illumination intensities and features to determine whether the captured image is from a live scene.

A method of processing image of a scene for liveness detection comprising the steps of detecting flicker fluctuations in the image captured by an image sensor by identifying a periodic pattern of illumination to determine inactivity of the scene, and analyzing the flicker occurrence by performing illumination analysis to determine whether the captured image is from a live scene.

Brief Description of the Drawings

Other objects, features, and advantages of the invention will be apparent from the following description when read with reference to the accompanying drawings. In the drawings, wherein like reference numerals denote corresponding parts throughout the several views:

Figure 1 shows block diagram of a liveness detection system in accordance to the present invention;

Figure 2 is a flowchart of the method of detecting flickering in fluorescent light source during periods of inactivity in an indoor scene;

Figure 3a is flowchart of the method of discriminating live scenes from fraudulent scenes based on the flicker of the light detected by a high speed camera; and Figure 3b is flowchart of the method of discriminating live scenes from fraudulent scenes based on the flicker of the light detected by a non high speed camera

Detailed Description of the Preferred Embodiments

In the following detailed description, numerous specific details are set forth in order to provide a thorough understanding of the invention. However, it will be understood by those or ordinary skill in the art that the invention may be practiced without these specific details. In other instances, well-known methods, procedures and/or components have not been described in detail so as not to obscure the invention. Reference will now be made in detail to the preferred embodiments of the present invention, examples of which are illustrated in the accompanying drawings.

With reference to Figure 1, a liveness-detection system (10) comprises an image sensor (11) for capturing images of a monitored area, a flicker detecting unit (12) to identify inactivity of the area and a flicker analyzing unit (13) for distinguishing flickering emanating from a live scene of the monitored area or that emanating from a photograph of the same scene. The flicker in the present invention refers to the flicker of fluorescent light source powered by an AC power supply where the flickering is the periodic brightness variations of the light source due to the cycling of the AC power supply.

A fluorescent light source is a gas-discharge light source that uses electricity to excite mercury vapor in argon or neon gas, resulting in a plasma that produces short-wave ultraviolet light. This light then causes a phosphor to fluoresce, producing visible light. The fluorescent light source will have a flicker frequency twice the frequency of the AC power main used to power the light source.

In the present invention, the term of fluorescent flicker used is not intended to limit the application of the system and method of the present invention to the detection of flicker generated by AC powered fluorescent light sources only. The liveness-detection system and method of the present invention can also be applied for the detection of flickering in images illuminated by any AC powered artificial light source that produces detectable flickering. The present invention also utilizes the image sensor to sample the overall illumination to detect whether the system operating in a 50 Hz, 60 Hz or incandescent light situation.

In some circumstances, fluorescent lamps operated at mains frequency can also produce flicker at the mains frequency (50 or 60 Hz) itself, which is noticeable by most people. This can also happen in the last few hours of tube life when the cathode emission coating at one end is almost run out, and that cathode starts having difficulty emitting enough electrons into the gas fill, resulting in slight rectification and hence uneven light output in positive and negative-going cycles of the AC power. Mains frequency flicker can also sometimes be emitted from the very ends of the tubes, as a result of each tube electrode alternately operating as an anode and cathode for each half mains cycle, and producing slightly different light output pattern in anode or cathode mode. Flicker at the mains frequency is more noticeable in peripheral vision than it is at the centre of gaze.

The image sensor (11) of a surveillance camera (not shown) views a monitored area which is illuminated by at least one fluorescent light (not shown) connected to the power source, whereby the light flickers at twice the power line frequency as the power being delivered to the light drops to zero twice per cycle. It is therefore, for example, if the power source has an output of a 60 Hz waveform, the fluorescent light will have intensity peaks of 120 Hz and on the other hand, if the power source has an output of a 50 Hz waveform, the intensity peaks of the lighting will be 100 Hz.

The image sensor (11) may captures images for the monitored area and the flickering of the fluorescent light at the monitored area is analyzed to distinguish a real live indoor scene from a fraudulent scene in which an equivalent photograph of the same scene is placed in front of the camera. Note that the photograph is assumed to be an exact replica of the scene to be precisely positioned in front of the camera such that there are neither edge misalignments nor intensity differences between the photograph and the real scene. Also note that the fluorescent light in the photograph does not flicker whereas that in a true live scene does.

The flicker detecting unit (12) that is used in the camera for photographing image at a frame rate and detects flicker of the image due to the light source whose brightness of which changes periodically. The flicker detection of the present invention is depending on the frame rate of the chosen camera. If high speed cameras are being used, for example having 2000 frames per second, then flicker can be identified by detecting a recurring pattern of global illumination changes. If non high-speed cameras are being used, for example having 50 frames per second, then flicker can be detected by the moving sequence of bands as the flicker causes dark bands across the captured image where these bands are usually referred to as the beat effect. Prior knowledge of these bands, coupled with basic motion analysis, are sufficient for efficient flicker detection here. The liveness-detection method starts by obtaining a plurality of successive live images from the image sensor (11) and the images are then transferred to a data unit. The illuminant intensities data from each successive live image are analyzed to obtain a live illumination data of the live images. Other possible statistical differences of flickering phenomena that holds between the live and fraudulent cases are analyzed by the flicker analyzing unit (13) . In the high speed camera situation, such differences may include the range of flicker induced global intensities and the local appearance of the flicker close to the fluorescent light itself or its photographic replica. Meanwhile, in the non high speed camera situation, such differences may include the differing contrast levels of the beat effect bands.

Images obtained by image acquisition (17) from the image sensor (11) are fed (14) into the flicker analyzing unit (13) when the system (10) detects (15) no activity after a certain length of time as shown in figure 2. Otherwise the system (10) will process (16) the surveillance images in the usual manner. In the high speed camera situation, the flicker analyzing unit (13) extracts (18) the global illumination data from a captured image as shown in figure 3a. The extracted data is then compared (19) with the live illumination data and the flicker analyzing unit (13) will classify (25) whether the captured scene is a live scene or a fraudulent scene. There is also an inclusion of the step of detecting (26) fluorescent light source in the high speed camera situation. This is an interesting problem by itself, since it is necessary to distinguish between the true light source and other objects that happen to be equally bright by virtue of their pigmentation and reflectance properties.

In the non high speed camera situation, the flicker analyzing unit (13) will first analyze (20) the motion detected in the captured image. The flicker analyzing unit (13) then extracts (21) bands and the band features (22) as shown in figure 3b. The band features are compared (23) with the live features and the flicker analyzing unit (13) will classify (24) whether the captured scene is a live scene or a fraudulent scene.

As will be readily apparent to those skilled in the art, the present invention may easily be produced in other specific forms without departing from its essential characteristics. The present embodiments is, therefore, to be considered as merely illustrative and not restrictive, the scope of the invention being indicated by the claims rather than the foregoing description, and all changes which come within therefore intended to be embraced therein.

Claims

1. A liveness detection system comprising: an image sensor for capturing a scene; a flicker detecting unit coupled to said image sensor that detects a periodic pattern of fluctuation in an image captured by said image sensor which indicate the presence of flicker; and > a flicker analyzing unit coupled to said image sensor and flicker detecting unit to perform illumination analysis on the illumination intensities and features to determine whether the captured image is of the live scene.

2. The liveness detection system as claimed in claim 1, wherein said system further comprising a data unit which is coupled to said image sensor for receiving successive captured images .

3. The liveness detection system as claimed in claim 1, wherein said image sensor is adapted to an image capturing apparatus.

4. The liveness detection system as claimed in claim 3, wherein said apparatus is a surveillance camera.

5. The liveness detection system as claimed in claim 1, wherein said image sensor is adapted to a high speed camera to detect flicker in the captured image.

6. The liveness detection system as claimed in claim 5, wherein said flicker is identified by detecting a repeating pattern of global illumination changes to identify inactivity in said scene over a period of time for flickering phenomena analysis .

7. The liveness detection system as claimed in claim 6, wherein said flickering phenomena analysis performed by said flicker analyzing unit to detect statistical differences between said live scene and fraudulent scenes.

8. The liveness detection system as claimed in claim 7, wherein in said fraudulent case, an equivalent photograph of the same scene is placed in front of said camera, where said photograph is assumed to be an exact replica of the scene and to be precisely positioned in such that there are neither edge misalignments nor intensity differences.

9. The liveness detection system as claimed in claim 7, wherein said analysis is performed to detect the range of flicker induced global intensities and the local appearance of the flicker close to the light source itself or its photographic replica.

10. The liveness detection system as claimed in claim 1, wherein said image sensor is adapted to a non high speed camera to detect flicker in the captured image.

11. The liveness detection system as claimed in claim 10, wherein said flicker is identified by detecting a moving sequence of dark bands to identify inactivity in said scene over a period of time for flickering phenomena analysis.

12. The liveness detection system as claimed in claim 11, wherein said flicker is further identified by performing a motion analysis prior to the bands detection.

13. The liveness detection system as claimed in claim 11, wherein said flickering phenomena analysis performed by said flicker analyzing unit to detect statistical differences between said live scene and fraudulent cases.

14. The liveness detection system as claimed in claim 13, wherein said analysis is performed to detect the differing contrast levels of the said dark bands.

15. The liveness detection system as claimed in claim 1, wherein said scene is an indoor scene.

16. A method of processing image of a scene for liveness detection comprising the steps of: detecting flicker fluctuation in said image captured by an image sensor by identifying a periodic pattern of illumination to determine inactivity of said scene; and analyzing said flicker occurrence by performing illumination analysis to determine whether the captured image is from a live scene.

17. The method of processing image as claimed in claim 16, wherein said step of detecting flicker includes the step of detecting a repeating pattern of global illumination changes for the image sensor which is adapted to a high speed camera.

18. The method of processing image as claimed in claim 16, wherein said step of analyzing flicker includes the step of performing the analysis of the statistical differences between said live scene and fraudulent scenes.

19. The method of processing image as claimed in claim 18, wherein said analysis is performed to detect the range of flicker induced global intensities and local appearance of the flicker close to the light source which is illuminating said scene itself or its photographic replica for the image sensor which is adapted to a high speed camera.

20. The method of processing image as claimed in claim 18, wherein said step of detecting flicker includes the step of detecting a moving sequence of dark bands in said image for the image sensor which is adapted to a non high speed camera.

21. The method of processing image as claimed in claim 18, wherein said analysis is performed to detect the differing contrast levels of the dark bands for the image sensor which is adapted to a non high speed camera.

22. The method of processing image as claimed in claim 20, wherein said prior to the step of detecting the dark bands, flicker is identified by performing a motion analysis to determine whether there is any flickering phenomenon.