A method for smoothing in-cabin liveness detection results, storage medium and electronic device

Abstract

The present invention provides a method for smoothing liveness detection results in a cabin, a storage medium and an electronic device, and belongs to the field of radar signal processing technology. The liveness detection result output adopts a gradually trusted strategy, and a detection flag is output in each frame. The detection flag has two states: the presence of a live body and the absence of a live body. The buffer area is stacked in units of a certain number of detection flags, and the result smoothing adopts a method of calculating the confidence of the detection flag buffer area, that is, setting a confidence threshold, and calculating the confidence of the buffer area after each detection flag stacking. The calculated confidence is compared with the confidence threshold to determine the liveness detection output result, and different process scheduling is completed according to the change of the cabin door state. The present invention can perform confidence analysis on the liveness detection results in the cabin, and then smooth the liveness detection results, which can overcome the shortcomings of underreporting and false alarms in the existing cabin liveness detection technology based on millimeter-wave radar, and can greatly improve the accuracy of cabin liveness detection.

Description

Technical Field

[0001] The present invention belongs to the field of radar signal processing, and in particular relates to a method for smoothing in-cabin living body detection results, a storage medium and an electronic device. Background Art

[0002] In recent years, due to the increase in the number of cars, new car evaluation regulations in various countries around the world have or are seeking to add in-car liveness detection items to the new regulations. At present, the more mature liveness detection technology is the in-car camera solution, but its privacy protection is poor. In response to this problem, based on the advantages of millimeter-wave radar such as good privacy protection, micro-Doppler feature sensitivity and low cost, FMCW millimeter-wave radar has become the mainstream choice for in-cabin liveness detection. Human breathing will modulate the radar echo. By demodulating the echo signal, the human breathing information can be extracted, which is used as the basis for liveness detection. However, the electromagnetic environment in the cabin is relatively complex. In addition to living organisms, there are clutter and other interferences, and it is impossible for living organisms to remain absolutely still. Some tiny displacements of living organisms and other interferences that do not affect the extraction of Doppler motion features but affect the extraction of breathing features may cause errors in the liveness detection results for a short period of time. Since the state of living organisms in the cabin only changes with the state of the cabin door, the state of living organisms in the cabin remains unchanged when the cabin door state remains unchanged, and the liveness detection results should not change. That is, the tiny missed alarms and false alarms caused by the interference mentioned above can be eliminated by appropriate technology. How to effectively eliminate short-term missed alarms and false alarms and reduce the missed alarm rate and false alarm rate of the liveness detection process has become an important research topic. Summary of the invention

[0003] The present invention aims at the problems of missed reports and false alarms in the prior art when detecting moving live objects in a millimeter-wave radar cabin caused by strong clutter interference from moving targets such as greatly shaking metal ornaments in the cabin, and provides a method for smoothing the results of live object detection in the cabin, a storage medium and an electronic device. The implementation method has a small amount of calculation and is easy to implement in hardware.

[0004] To achieve the above object, the present invention adopts the following technical solutions:

[0005] A method for smoothing in-cabin liveness detection results, characterized in that it comprises the following steps:

[0006] Step 1: Set a result smoothing buffer area to store the liveness detection flag obtained for each frame of the liveness detection process;

[0007] Step 2: Set the parameters of the result smoothing buffer area to zero and clear the result smoothing buffer area;

[0008] Step 3: stacking the liveness detection marks obtained for each frame in the liveness detection process;

[0009] Step 4: After a certain number of liveness detection marks are stacked, the confidence of the result smoothing buffer area is calculated;

[0010] Step 5: Set a confidence threshold, and determine the relationship between the confidence of the result smoothing buffer and the threshold. If the confidence of the result smoothing buffer is greater than the threshold, output the confidence result as the in-cabin liveness detection result, otherwise repeat steps 3 and 4;

[0011] Step 6: Continue to detect the door status mark. If the door status changes, restart the process from step 2.

[0012] To optimize the above technical solutions, the specific measures taken also include:

[0013] Furthermore, in step 1, while setting the result smoothing buffer area, the number of stacking times M and the buffer area confidence CL after the Mth stacking are set. M .

[0014] Furthermore, in step 2, the parameters of the result smoothing buffer area are set to zero, specifically, the number of stacking times M is set to 0, and the confidence level CL of the result smoothing buffer area is set to 0.

[0015] Furthermore, in step 3, the liveness detection process will obtain a Boolean liveness detection flag for each frame, as follows:

[0016]

[0017] The liveness detection flag obtained for each frame of the liveness detection process is stacked in units of N frames and stored in a result smoothing buffer area, and each time N frames are stacked, the number of stacking times is M=M+1.

[0018] Furthermore, in step 4, the confidence calculation of the result smoothing buffer area is specifically as follows:

[0019] After M stacking, there are M×N liveness detection marks in the result smoothing buffer area. Calculate the confidence CL in the result smoothing buffer area after the Mth stacking. M :

[0020]

[0021] In the formula, K true and K false They are the number of liveness detection flags with values ​​of True and False in the result smoothing buffer at this time, respectively.

[0022] Furthermore, in step 5, a confidence threshold Threshold is set, and the confidence CL obtained in step 4 is MCompare with Threshold and decide the next steps of the smoothing process based on the comparison result:

[0023] If CL M >Threshold, it is considered that the living state in the cabin has been determined, and the subsequent output results are always confidence results. The confidence results are the living body detection flags that account for a large proportion in the result smoothing buffer area, as follows:

[0024]

[0025] If CL M ≤Threshold, the status of the living organisms in the cabin is still uncertain, and steps 3 and 4 are repeated until CL M >Threshold, the output result of the process is the untrusted result of each frame.

[0026] Further, in step 6, the door status flag detection occurs at CL M ≤Threshold, if the door status is detected to have changed, the process will restart from step 2; otherwise, steps 3 and 4 will be repeated until CL M >Threshold.

[0027] Further, in step 6, the door status flag detection occurs at CL M >Threshold, if the door status is detected to have changed, the process will be restarted from step 2 until CL M >Threshold; otherwise continue to output confidence results.

[0028] On the other hand, the present invention further proposes a computer-readable storage medium storing a computer program, characterized in that the computer program enables a computer to execute the in-cabin living body detection result smoothing method as described above.

[0029] On the other hand, the present invention also proposes an electronic device, characterized in that it includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for smoothing the in-cabin living body detection results as described above is implemented.

[0030] The beneficial effects of the present invention are as follows: the present invention is based on the radar signal processing steps, and the output of the liveness detection result adopts a "gradual belief" strategy. The detection flag is output in each frame. The detection flag has two states: the presence and the absence of a live body. The cache area is stacked in units of a certain number of detection flags. The result is smoothed by using a method for calculating the confidence of the detection flag cache area, that is, a confidence threshold is set. After each detection flag stacking, the cache area confidence calculation is performed. The calculated confidence is compared with the confidence threshold to determine the liveness detection output result, and different process scheduling is completed according to the change of the cabin door state.

[0031] The "gradual belief" result smoothing strategy adopted by the present invention effectively avoids the problem of missed detection results in a short period of time caused by tiny displacements of living bodies and other interferences that do not affect the extraction of Doppler motion features but affect the extraction of respiratory features, and the missed detection rate can be 0 after the result is confident; the method of setting the confidence threshold adopted by the present invention effectively avoids the problem of false alarms in detection results caused by certain tiny interferences when there are no living bodies in the cabin, and the false alarm rate can be 0 after the result is confident; the present invention also continues to perform cabin door status mark detection during the smoothing process, effectively avoiding errors caused by changes in cabin door status, and further improving the accuracy of cabin live detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0032] Figure 1 This is a flow chart of the method for smoothing the results of in-cabin liveness detection.

[0033] Figure 2 Schematic diagram of the in-cabin liveness detection marker stacking and confidence calculation method.

[0034] Figure 3 Schematic diagram of the smoothing results of liveness detection in a typical scenario. DETAILED DESCRIPTION

[0035] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0036] The idea of ​​the present invention is based on the radar signal processing steps. The output of the liveness detection result adopts the strategy of "gradual belief". The detection flag is output in each frame. The detection flag has two states: the presence and absence of liveness. The buffer area is stacked with a certain number of detection flags as units. The result smoothing adopts the method of detection flag buffer area confidence calculation, that is, setting a confidence threshold. After each detection flag stacking, the cache area confidence calculation is performed. The calculated confidence is compared with the confidence threshold to determine the liveness detection output result, and different process scheduling is completed according to the change of the cabin door state.

[0037] In one embodiment, if Figure 1 As shown, the present invention provides a method for smoothing the results of in-cabin liveness detection. The method is divided into four steps: pre-parameter setting, detection flag stacking, confidence calculation and analysis, and cabin door status process scheduling.

[0038] Step 1: Pre-parameter setting.

[0039] First, set the result smoothing buffer to store the liveness detection flag of each frame, set the number of buffer stacking times M, and the buffer confidence CL after the Mth stacking M , set both parameters to 0, that is, M = 0, CL = 0, clear the result smoothing buffer, and set the subsequent confidence judgment threshold Threshold.

[0040] Step 2: Check flag stacking.

[0041] Figure 2 The following is a schematic diagram of the stacking of in-cabin liveness detection marks and the confidence calculation method. The radar liveness detection process will obtain a detection mark for each frame. The detection mark is a Boolean value, as follows:

[0042]

[0043] The liveness detection flags obtained in each frame liveness detection process are stacked in units of N frames and stored in a buffer area, and the number of stacking times for each stack of N frames is M=M+1.

[0044] Step 3: Confidence calculation and analysis.

[0045] After M stacking, there are M×N liveness detection marks in the result smoothing buffer area. Calculate the confidence of the result in the buffer area after the Mth stacking:

[0046]

[0047] In the formula, K true and K false They are the number of flags with values ​​of True and False in the buffer at this time, respectively.

[0048] The confidence level CL obtained aboveM Compared with the confidence threshold Threshold, the next step of the smoothing process is determined based on the comparison result. If CL M >Threshold, it is considered that the living state in the cabin has been determined, and the subsequent output results are always the confidence results, specifically the detection flags that account for a large proportion in the result smoothing buffer area, that is,

[0049]

[0050] If CL M ≤Threshold, it is considered that the living state in the cabin is still undetermined, and the above detection mark stacking and confidence calculation steps are repeated until CL M >Threshold (M will not exceed 8 in the measured data), and the output result in the process is the untrusted result of each frame.

[0051] Step 4: Door status process scheduling.

[0052] A prerequisite for in-cabin liveness detection is that the in-cabin liveness status changes only with the opening and closing of the cabin door. Therefore, the cabin door status is continuously monitored during the result smoothing process. If the cabin door status changes, the previously obtained stacking times, confidence, and buffer area are cleared, and the detection mark stacking and confidence calculation steps are continued.

[0053] Door status sign detection can be divided into two types according to the timeline, specifically:

[0054] (1) The door status flag detection occurs in CL M ≤Threshold, that is, during the process of stacking the detection marks and calculating the confidence, if the door state is detected to have changed, the stacking times M is set to 0, and the buffer area is cleared, and the detection mark stacking and confidence calculation steps are continued; otherwise, the detection mark stacking and confidence calculation steps are directly performed until CL M >Threshold.

[0055] (2) The door status flag detection occurs at CL M >Threshold, it is considered that the state of the living body in the cabin has been determined. If the cabin door state is detected to have changed, the stacking times M is set to 0, and the buffer area is cleared, and the detection mark stacking and confidence calculation steps are performed until CL M >Threshold; otherwise continue to output the confident result.

[0056] The above four steps complete the smoothing process of in-cabin liveness detection results.

[0057] Figure 3This is a schematic diagram of the smoothing results of liveness detection in a typical scenario in the experiment. The experimental scenario is in the car cabin, the radar is installed on the top of the rear row, and the radar configuration parameters are: starting frequency is 60GHz, bandwidth is 4GHz, ADC sampling rate is 20Msps, frequency modulation period and frame period are 25.6us and 200ms respectively. The liveness item uses a simulation doll to simulate a real live target, which is placed on the left foot pad about 1m away from the radar, and an umbrella is placed above the simulation doll. The test duration is 650s. The method of the present invention is used to smooth the liveness detection result. The liveness detection result before smoothing is Figure 3 In the figure above, the smoothed liveness detection result is Figure 3 As shown in the figure below, during the experimental time, the unsmoothed detection results have missed detections, while the missed detection rate of the liveness detection results after using the method of the present invention is 0, which indicates that the method of the present invention has significantly improved the accuracy of the liveness detection in the cabin.

[0058] In another embodiment, the present invention provides a computer-readable storage medium storing a computer program, wherein the computer program enables a computer to execute the method for smoothing in-cabin living body detection results as described in the first embodiment.

[0059] In another embodiment, the present invention proposes an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the method for smoothing the in-cabin living body detection results as described in the first embodiment is implemented.

[0060] In the embodiments disclosed in the present application, the computer storage medium can be a tangible medium that can contain or store a program for use by or in conjunction with an instruction execution system, device or equipment. The computer storage medium can include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices or equipment, or any suitable combination of the above. More specific examples of computer storage media can include electrical connections based on one or more lines, portable computer disks, hard disks, random access memories (RAM), read-only memories (ROM), erasable programmable read-only memories (EPROM or flash memory), optical fibers, portable compact disk read-only memories (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the above.

[0061] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed in this application can be implemented in electronic hardware or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0062] The above are only preferred embodiments of the present invention, and the protection scope of the present invention is not limited to the above embodiments. All technical solutions under the concept of the present invention belong to the protection scope of the present invention. It should be pointed out that for ordinary technicians in this technical field, some improvements and modifications without departing from the principle of the present invention should be regarded as the protection scope of the present invention.

Claims

1. A method for smoothing in-cabin liveness detection results, characterized in that: The steps include: Step 1: Set a result smoothing buffer area to store the liveness detection flag obtained for each frame of the liveness detection process; Step 2: Set the parameters of the result smoothing buffer area to zero and clear the result smoothing buffer area; Step 3: stacking the liveness detection flags obtained for each frame of the liveness detection process; the liveness detection process will obtain a Boolean liveness detection flag for each frame, as follows: The liveness detection flags obtained from each frame of the liveness detection process are stacked in units of N frames and stored in the result smoothing buffer area, and each time N frames are stacked, the number of stacking times is M=M+1; Step 4: After a certain number of liveness detection marks are stacked, the confidence calculation of the result smoothing buffer area is performed; the confidence calculation of the result smoothing buffer area is specifically as follows: After M stacking, there are M×N liveness detection marks in the result smoothing buffer area. Calculate the confidence CL in the result smoothing buffer area after the Mth stacking. M : In the formula, K true and K false are the number of liveness detection flags with values ​​of True and False in the result smoothing buffer at this time, respectively; Step 5: Set the confidence threshold, determine the relationship between the confidence of the result smoothing buffer and the threshold, if the confidence of the result smoothing buffer is greater than the threshold, then output the confidence result as the in-cabin living body detection result, otherwise repeat steps 3 and 4; specifically: set the confidence threshold Threshold, and convert the confidence CL obtained in step 4 M Compare with Threshold and decide the next steps of the smoothing process based on the comparison result: If CL M >Threshold, it is considered that the living state in the cabin has been determined, and the subsequent output results are always confidence results. The confidence results are the living body detection flags that account for a large proportion in the result smoothing buffer area, as follows: If CL M ≤Threshold, the status of the living organisms in the cabin is still uncertain, and steps 3 and 4 are repeated until CL M >Threshold, the output result of the process is the untrusted result of each frame; Step 6: Continue to detect the door status mark. If the door status changes, restart the process from step 2. Specifically: the door status mark detection occurs at CL M ≤Threshold, if the door status is detected to have changed, the process will restart from step 2; otherwise, steps 3 and 4 will be repeated until CL M >Threshold; the door status flag detection occurs at CL M >Threshold, if the door status is detected to have changed, the process will be restarted from step 2 until CL M >Threshold; otherwise continue to output confidence results.

2. The method for smoothing in-cabin liveness detection results according to claim 1, characterized in that: In step 1, while setting the result smoothing buffer area, the number of stacking times M and the confidence level CL of the buffer area after the Mth stacking are set. M .

3. The method for smoothing in-cabin liveness detection results according to claim 1, characterized in that: In step 2, the parameters of the result smoothing buffer area are set to zero, specifically: the number of stacking times M is set to 0, and the confidence level CL of the result smoothing buffer area is set to 0.

4. A computer-readable storage medium storing a computer program, characterized in that: The computer program enables a computer to execute the method for smoothing in-cabin living body detection results as described in any one of claims 1 to 3.

5. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the method for smoothing the in-cabin living body detection result as described in any one of claims 1 to 3 is implemented.

Images