Liveness detection method and related device

Abstract

The present invention discloses a living body detection method and related devices, including: obtaining at least one complete cycle of rPPG signal from an image containing a skin area, extracting an rPPG waveform feature from the rPPG signal, and judging whether there is a living body in the image according to the rPPG waveform feature. The computing device includes a processing unit for executing program code, a storage unit connected to the processing unit and for storing program code; the program code instructs the processing unit to execute the aforementioned method. A living body detection device includes a signal extraction unit for receiving an image containing a skin area and obtaining at least one complete cycle of rPPG signal from the image; a waveform feature extraction unit for extracting an rPPG waveform feature according to the rPPG signal; and a judgment unit for judging whether there is a living body in the image according to the rPPG waveform feature.

Description

Technical Field

[0001] The present invention relates to a living body detection method, in particular to a method and a related device for identifying whether the skin in an image is a living body or a false image by remote photoplethysmography. Background Art

[0002] Traditional mechanisms for implementing liveness detection include using color texture, image reflectivity, and image depth information in images to prevent malicious people from deceiving the face recognition system through printed paper, photos, videos, or 3D masks. However, these mechanisms cannot effectively identify high-quality or high-resolution deception methods. There are also mechanisms that require the person to be identified to make active responses such as blinking or opening the mouth, but this method of requiring the person to be identified to make active responses often results in a poor user experience.

[0003] There is also a mechanism for using physiological signals to achieve liveness detection. This mechanism is mainly based on remote photoplethysmography (rPPG). Remote photoplethysmography is a method that uses a camera to detect human heartbeat waveforms and estimate heart rate in a non-contact manner, so it can determine whether the face in the image is a real face. In detail, the human heart activity generates physiological signals in the body, that is, changes in the amount of blood congestion in the subcutaneous microvessels, which causes changes in the blood's absorption rate of light, so it can monitor changes in skin brightness caused by blood flow. The concept of remote photoplethysmography is to detect physiological signals (referred to as rPPG signals in this article) by analyzing the color changes of the skin area. Simply put, rPPG signals can reflect a variety of physiological information, such as changes in the heart cycle, changes in blood volume in blood vessels, heart rate values, etc. However, the main difficulties of this technology include light source changes and motion interference.

[0004] For example, the existing method of using rPPG signals to determine whether a person is alive requires obtaining the rPPG spectrum through time-frequency conversion. However, if the false image shakes at a fixed frequency or the fixed frequency light source changes, a periodic signal will be generated. Therefore, the physiological signals calculated from this periodic signal, such as heart rate, are easily controlled within the normal heart rate range, resulting in the misjudgment of the presence of a living person in the image. In addition, the time-frequency conversion algorithm requires a large number of rPPG signals to be calculated, so the rPPG spectrum will not be stable until a considerable number of signals are collected, resulting in low recognition efficiency of the subsequent face recognition system.

[0005] In addition, the traditional rPPG spectrum obtained after time-frequency conversion further performs spectrum disorder calculation to determine whether there is a living body. However, in the case of fixed frequency shaking or fixed frequency light source changes, a periodic signal will still be generated, so the spectrum disorder is also easily within the qualified range, resulting in misjudgment of living bodies. In addition, through multiple rPPG signals obtained in different intervals of interest, cross-correlation waveform calculation is performed to obtain the waveform similarity of the two regions. However, similarly, this method still has the above-mentioned problems.

[0006] The disclosure of the above background technology content is only used to assist in understanding the inventive concept and technical solution of the present invention. It does not necessarily belong to the prior art of this patent application. In the absence of clear evidence that the above content has been disclosed before the filing date of this patent application, the above background technology should not be used to evaluate the novelty and creativity of this application. Summary of the invention

[0007] The main purpose of the present invention is to provide a living body detection method and related devices, which can improve the misjudgment of living bodies caused by light source changes and motion interference, so as to solve the above-mentioned problems.

[0008] The present invention proposes the following technical solutions to achieve the above-mentioned purpose:

[0009] A method for detecting a living body, used in a computing device, comprises: obtaining at least one complete cycle of an rPPG signal from an image containing a skin area; extracting an rPPG waveform feature from the rPPG signal; and determining whether a living body exists in the image based on the rPPG waveform feature.

[0010] A computing device comprises: a processing unit for executing a program code; a storage unit connected to the processing unit for storing the program code; wherein the program code instructs the processing unit to execute the following steps: obtaining at least one complete cycle of rPPG signal from a skin image; extracting an rPPG waveform feature from the rPPG signal; and determining whether there is a living body in the skin image based on the rPPG waveform feature.

[0011] A living body detection device comprises: a signal acquisition unit for receiving an image containing a skin area and obtaining at least one complete cycle of rPPG signal from the image; a waveform feature acquisition unit for acquiring an rPPG waveform feature according to the rPPG signal; and a judgment unit for judging whether a living body exists in the image according to the rPPG waveform feature. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 This is a schematic diagram of a living body detection device according to a first embodiment of the present invention;

[0013] Figure 2 This is a flow chart of a living body detection method according to a first embodiment of the present invention;

[0014] Figure 3 Schematic diagram of waveform characteristics of the first embodiment of the present invention;

[0015] Figure 4 FIG. 1 is a schematic diagram of the operation of a living body detection device according to an embodiment of the present invention.

[0016] Description of reference numerals:

[0017] 10.40 Liveness detection device

[0018] 100 Processing Units

[0019] 110 Storage Units

[0020] 120 Communication interface unit

[0021] 114 Program code

[0022] 20 Process

[0023] Steps 201-203

[0024] 410 Signal Acquisition Unit

[0025] 420 Waveform feature extraction unit

[0026] 430 Judgment Unit DETAILED DESCRIPTION

[0027] The present invention will be further described below in conjunction with the accompanying drawings and specific implementation methods.

[0028] Please refer to Figure 1 , Figure 1Schematic diagram of a liveness detection device 10 according to an embodiment of the present invention. The liveness detection device 10 may be a local computing device or a cloud device, and includes a processing unit 100, a storage unit 110, and a communication interface unit 120. The processing unit 100 may be a microprocessor or an application-specific integrated circuit (ASIC). The storage unit 110 may be any data storage device for storing a program code 114, and the program code 114 is read and executed by the processing unit 100. For example, the storage unit 110 may be a subscriber identity module (SIM), a read-only memory (ROM), a random-access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, etc., but is not limited thereto. The communication interface unit 120 may transmit signals between other devices (such as a camera and / or a display device) by wired or wireless communication. For example, the communication interface unit 120 can receive an image from a camera and transmit it to the processing unit 100, and the result generated by the processing unit 100 can be further transmitted to the display device. Therefore, the display device can indicate the detection result of the living body detection device 10, such as whether there is a real living body in the image, or it is a false image without a real living body, and then notify the security system or generate an alarm.

[0029] Please refer to Figure 2 , which is a schematic diagram of a liveness detection process 20 according to an embodiment of the present invention. The liveness detection process 20 is used to Figure 1 The living body detection device 10 shown in the figure can be compiled into a program code 114 and includes the following steps 201-203:

[0030] Step 201: Obtain at least one complete cycle of rPPG signal from a skin image.

[0031] Step 202: Extract rPPG waveform features from the complete cycle of the rPPG signal.

[0032] Step 203: Determine whether the skin image contains a living body based on the rPPG waveform characteristics.

[0033] From the above, it can be seen that the present invention proposes to capture a complete rPPG cycle signal and use the rPPG waveform characteristics to determine whether there is a living body. Figure 3, which is a schematic diagram of the waveform characteristics of the first embodiment of the present invention. In detail, the rPPG waveform characteristics include the arterial notch waveform (dicrotic notch), systolic pressure waveform (systolic), diastolic pressure waveform (diastolic), systolic-diastolic pressure waveform (systolic-diastolic), rPPG first-order differential characteristics (such as Figure 3 The waveform in the middle) and the second-order differential characteristics of rPPG (such as Figure 3 Further, the rPPG second-order differential feature includes an early systolic positive wave (early systolic positive wave, Figure 3 The early systolic negative wave (marked as a) in the middle Figure 3 b), late systolic reincreasing wave Figure 3 c), late systolic redecreasing wave, Figure 3 The early diastolic positive wave (marked as d) Figure 3 In other words, when the rPPG waveform features include at least one of the arterial notch waveform, the systolic pressure waveform, the diastolic pressure waveform, the systolic and diastolic pressure waveform, the rPPG first-order differential feature, and the rPPG second-order differential feature, the living body detection device 10 determines that there is a living body in the skin image. On the contrary, if the rPPG waveform features do not include the above waveforms, the skin image is determined to be a false image.

[0034] In one embodiment, the living body detection device 10 may perform waveform-related operations or algorithms on the rPPG signal, such as signal amplification or noise filtering, to enhance the rPPG waveform characteristics.

[0035] See also Figure 4, which is an operation diagram of a liveness detection device according to an embodiment of the present invention. The liveness detection device 40 includes a signal acquisition unit 410, a waveform feature acquisition unit 420, and a judgment unit 430. The signal acquisition unit 410 is used to obtain at least one complete cycle of rPPG signals from the skin image, and then transmit the rPPG cycle signal to the waveform feature acquisition unit 420. The waveform feature acquisition unit 420 is used to obtain rPPG waveform features from the rPPG cycle signal, and then transmit it to the judgment unit 430. The judgment unit 440 is used to determine whether there is a real live body in the image according to the rPPG waveform features and a preset threshold. In one embodiment, the judgment unit 430 can determine the probability of waveform features and / or adjust the threshold according to the mathematical template obtained by database statistics or machine learning technology, and then determine whether the rPPG waveform features exist. For example, if the rPPG waveform feature parameter is higher than the threshold, the judgment unit 430 determines that the image contains a live body; on the contrary, if it is lower than the threshold, it is determined that the image does not contain a live real person image.

[0036] All the above steps, including the suggested steps, can be implemented by hardware, software (i.e., a combination of hardware devices and computer instructions, where the information in the hardware devices is read-only software information) or electronic systems. For example, the hardware may include analog circuits, digital circuits and hybrid circuits (i.e., microcircuits, microchips or silicon chips). The electronic system may include a system on chip (SOC), a system in package (SIP), a computer on module (COM) and a liveness detection device 10, 40.

[0037] In summary, the liveness detection method and related device provided by the present invention utilizes rPPG periodic signals to obtain its waveform characteristics to directly determine whether there is a live body, without performing time-frequency conversion operations, thereby avoiding misjudgment of liveness caused by regular frequency fluctuations or changes in fixed frequency light sources. Therefore, compared with traditional methods, the present invention has a higher accuracy of liveness detection performance.

[0038] The above contents are further detailed descriptions of the present invention in combination with specific preferred embodiments, and it cannot be determined that the specific implementation of the present invention is limited to these descriptions. For those skilled in the art of the present invention, several equivalent substitutions or obvious variations can be made without departing from the concept of the present invention, and the performance or use is the same, which should be regarded as belonging to the protection scope of the present invention.

Claims

1. A method for detecting a living body, used in a computing device, characterized in that: include: Acquire at least one complete cycle of rPPG signal from an image containing a skin area; Extracting an rPPG waveform feature from the rPPG signal; as well as, Determining whether there is a living body in the image according to the rPPG waveform characteristics; Wherein, the step of judging whether there is a living body in the image according to the rPPG waveform characteristics comprises: According to the mathematical template obtained from the database statistics, or the machine learning technology, the probability of the waveform feature appearing and / or the threshold value are determined, and then the existence of the rPPG waveform feature is determined; when the rPPG waveform feature includes at least one of the arterial notch waveform, the systolic pressure waveform, the diastolic pressure waveform, the systolic and diastolic pressure waveform, the rPPG first-order differential feature and the rPPG second-order differential feature, it is determined that there is a living body in the image; otherwise, the image is determined to be a false image.

2. The method for detecting a living body according to claim 1, wherein: The rPPG second-order differential feature includes at least one of an early systolic positive wave, an early systolic negative wave, a late systolic re-increase wave, a late systolic re-decrease wave and an early diastolic positive wave.

3. A computing device, characterized in that: Include: a processing unit for executing a program code; a storage unit, connected to the processing unit, for storing the program code; The program code instructs the processing unit to execute the following steps: Obtain at least one complete cycle of rPPG signal from a skin image; Extracting an rPPG waveform feature from the rPPG signal; and Determining whether there is a living body in the skin image according to the rPPG waveform characteristics; Wherein, the step of judging whether there is a living body in the image according to the rPPG waveform characteristics comprises: According to the mathematical template obtained from the database statistics, or the machine learning technology, the probability of the waveform feature appearing and / or the threshold value are determined, and then the existence of the rPPG waveform feature is determined; when the rPPG waveform feature includes at least one of the arterial notch waveform, the systolic pressure waveform, the diastolic pressure waveform, the systolic and diastolic pressure waveform, the rPPG first-order differential feature and the rPPG second-order differential feature, it is determined that there is a living body in the image; otherwise, the image is determined to be a false image.

4. The computing device according to claim 3, wherein: The rPPG second-order differential feature includes at least one of an early systolic positive wave, an early systolic negative wave, a late systolic re-increase wave, a late systolic re-decrease wave and an early diastolic positive wave.

5. A living body detection device, characterized in that: Contains: a signal acquisition unit, configured to receive an image including a skin area, and obtain at least one complete cycle of rPPG signal from the image; a waveform feature extraction unit, used for extracting an rPPG waveform feature according to the rPPG signal; and a judgment unit, configured to judge whether there is a living body in the image according to the rPPG waveform characteristics; The judgment unit judges the probability of occurrence of waveform features and / or adjusts the threshold value according to the mathematical template obtained by database statistics or the machine learning technology, and further judges whether the rPPG waveform features exist, and judges that the living body exists in the image when it is judged that the rPPG waveform features include at least one of the arterial notch waveform, the systolic pressure waveform, the diastolic pressure waveform, the systolic and diastolic pressure waveform, the rPPG first-order differential feature and the rPPG second-order differential feature; and judges that the image is a false image when it is judged that the rPPG waveform features do not include the arterial notch waveform, the systolic pressure waveform, the diastolic pressure waveform, the systolic and diastolic pressure waveform, the rPPG first-order differential feature and the rPPG second-order differential feature.

6. The living body detection device according to claim 5, characterized in that: The rPPG second-order differential feature includes at least one of an early systolic positive wave, an early systolic negative wave, a late systolic re-increase wave, a late systolic re-decrease wave and an early diastolic positive wave.

Images