Data processing method and data processing system

Abstract

To provide a data processing method and a data processing system capable of improving accuracy of living body detection in a vehicle.SOLUTION: A data processing method for processing data collected by a millimeter-wave radar, comprising: collecting in-vehicle actual measurement data; and removing noise in the in-vehicle actual measurement data based on a noise map, wherein the noise map is generated based on in-vehicle environment data collected when it is determined that no living body exists in a vehicle.SELECTED DRAWING: Figure 3

Description

[Technical Field]

[0001] The present invention relates to a data processing method and a data processing system. [Background technology]

[0002] In the field of automotive safety, detecting living bodies such as children and pets in a vehicle is an important research topic. To achieve this, occupant detection technology based on millimeter-wave radar is typically used. Millimeter-wave radar is a radar device that emits discontinuous radar signals and analyzes the returned radar signals to detect the presence, distance, speed, etc. of a target. Specifically, in millimeter-wave radar-based occupant detection technology, after the driver leaves the vehicle, the millimeter-wave radar emits millimeter-wave radar signals within a short period (e.g., 10 seconds), analyzes the collected radar signals for life signatures, and thereby identifies living bodies (e.g., children, sleeping adults, pets, etc.) in the vehicle. When a living body is identified, an alarm is sounded, thereby ensuring the safety of the vehicle occupants. Furthermore, conventional millimeter-wave radar-based occupant detection technology requires processing the collected radar signals in a manner that reduces the influence of noise present in the vehicle environment before analyzing and identifying the radar signals.

[0003] Common data processing methods for noise reduction include real-time data processing methods such as signal strength enhancement, resolution improvement, circuit filtering, software filtering, etc. However, since the collected radar signal simultaneously contains signals from targets (e.g., living organisms) and signals from noise, the conventional data processing methods for noise reduction have various problems as follows:

[0004] First, in conventional data processing methods, when noise is reduced based on noise separation parameters, the collected radar signals simultaneously contain signals from targets and noise, resulting in low reliability of the noise separation parameters. Meanwhile, obtaining reliable noise separation parameters requires highly complex calculation methods, resulting in increased maintenance costs and reduced processing efficiency. Second, in conventional data processing methods, in order to achieve real-time noise reduction, the accuracy of the algorithm used in the data processing method is directly proportional to the processing time, making it impossible to achieve both algorithm accuracy and processing time. When implementing such data processing methods, achieving sufficiently high algorithm accuracy requires lowering performance and increasing processing time, resulting in delayed alerts and potentially late rescue. Finally, in conventional data processing methods, each time a living organism is detected inside a vehicle, historical data from the previous detection is discarded and new data is obtained, which prevents full utilization of idle resources such as sensors, further limiting the performance and processing efficiency of the data processing method.

[0005] Therefore, when detecting a living body inside a vehicle using conventional data processing methods, it is particularly susceptible to noise in the vehicle environment, resulting in an increase in false detections, a decrease in detection accuracy, and an enormous amount of data to be processed.

[0006] In another conventional data processing method, noise in the collected radar signal is removed each time a vehicle interior is detected. However, the interior environment changes over time, for example, as more items are placed inside the vehicle. Therefore, using the data representing the interior environment at the time of shipment, it is not possible to accurately remove noise, resulting in reduced detection accuracy. [Prior art documents] [Patent documents]

[0007] [Patent Document 1] Japanese Patent Application Laid-Open No. 2000-338231 [Patent Document 2] Japanese Patent Application Laid-Open No. 2011-209238 [Patent Document 3] Japanese Patent Publication No. 2020-012683 Summary of the Invention [Problem to be solved by the invention]

[0008] The present invention has been made in view of the above technical problems, and has an object to provide a data processing method and a data processing system that can improve the accuracy of detecting living bodies inside a vehicle. [Means for solving the problem]

[0009] A data processing method according to the present invention processes data collected using a millimeter-wave radar and has the following features: It includes the steps of collecting in-vehicle actual measurement data and removing noise from the in-vehicle actual measurement data based on a noise map, where the noise map is generated based on in-vehicle environment data collected when it is confirmed that no living body is present in the vehicle.

[0010] Thus, when it is confirmed that no living body is present in the vehicle, the data processing method of the present invention collects in-vehicle environment data, generates a noise map, and removes noise from the in-vehicle actual measurement data based on the noise map. In this way, when a living body is detected in the vehicle by analyzing the in-vehicle actual measurement data from which noise has been removed, the accuracy of detecting a living body in the vehicle can be improved.

[0011] Furthermore, the data processing method of the present invention generates a noise map using the vehicle interior environment data collected when it is confirmed that no living organism is present in the vehicle, and processes the vehicle interior measured data. Therefore, compared to a method of removing noise using fixed, preset vehicle interior environment data at the time of shipment, noise can be removed more accurately, thereby improving the accuracy of detecting a living organism in the vehicle.

[0012] A data processing method according to the present invention is characterized in that the in-vehicle environment data includes at least one of time domain data, frequency domain data, and spatial data.

[0013] As a result, the data processing method of the present invention converts the in-vehicle environment data collected when it is confirmed that no living organism is present inside the vehicle into at least one of three-dimensional data such as time domain data, frequency domain data, and spatial data. Therefore, information such as the time, type, source, and location of noise is identified in three dimensions. This allows for more accurate noise removal, improving the accuracy of in-vehicle living organism detection. It also reduces the amount of calculation required to detect a living organism inside the vehicle, improving processing efficiency.

[0014] In the data processing method according to the present invention, in the step of removing noise from the in-vehicle actual measurement data, noise information in the in-vehicle actual measurement data is identified based on the noise map, the time domain data, the frequency domain data, and the spatial data, and the noise is removed using a noise reduction method corresponding to the noise information.

[0015] As a result, the data processing method of the present invention identifies noise information such as the time, type, source, and location of noise in three dimensions, such as time domain data, frequency domain data, and spatial data. This allows appropriate noise reduction methods to be used depending on the different noise conditions indicated by the noise information, allowing for more accurate noise removal and improving the accuracy of in-vehicle life detection. It also improves processing efficiency.

[0016] The data processing method according to the present invention is characterized in that in the step of removing noise from the in-vehicle actual measurement data, data within the time range in the in-vehicle actual measurement data is removed based on information indicating a time range in which the noise occurs in the time domain data.

[0017] The data processing method according to the present invention is characterized in that in the step of removing noise from the in-vehicle actual measurement data, data in the frequency range from the in-vehicle actual measurement data is removed based on information indicating a frequency range of the noise in the frequency domain data.

[0018] The data processing method according to the present invention is characterized in that in the step of removing noise from the in-vehicle actual measurement data, data in the position range in the in-vehicle actual measurement data is removed based on information indicating the position range of occurrence of the noise in the spatial data.

[0019] As a result, the data processing method according to the present invention removes noise in three dimensions, such as time domain data, frequency domain data, and spatial data. Furthermore, data processing is performed in a manner that removes noise from in-vehicle measured data for detecting a living body inside the vehicle. This allows for more accurate noise removal, improving the accuracy of in-vehicle living body detection. Furthermore, because unnecessary data is removed, the amount of calculation required for detecting a living body inside the vehicle can be reduced, improving processing efficiency.

[0020] The data processing method according to the present invention is characterized in that in the step of confirming that no living body is present inside the vehicle, it is confirmed that the living body is not present inside the vehicle based on an operation by a user.

[0021] The data processing method according to the present invention is characterized in that a characteristic portion of the in-vehicle actual measurement data is amplified by a filtering algorithm.

[0022] A data processing system according to the present invention processes data collected using a millimeter wave radar, and has the following features: The data processing system uses the data processing method according to any one of claims 1 to 7. [Effects of the Invention]

[0023] The data processing method and system according to the present invention can improve the accuracy of in-vehicle live body detection.

[0024] Furthermore, the data processing method and system according to the present invention can accurately remove noise, reduce the amount of calculation required to detect a living body inside a vehicle, and improve processing efficiency. [Brief explanation of the drawings]

[0025] [Figure 1] FIG. 1 is a block diagram illustrating an example of the configuration of a data processing system. [Figure 2A] FIG. 1 is a block diagram showing an example of the configuration of a millimeter wave radar. [Figure 2B] FIG. 1 is a schematic diagram showing the spatial distribution of millimeter wave radar within a vehicle cabin. [Figure 3] 1 is a flowchart illustrating a process for detecting a living body in a vehicle. [Figure 4] 10 is a flowchart illustrating noise map generation. [Figure 5] FIG. 10 is a schematic diagram showing a process of removing noise from in-vehicle actual measurement data based on a noise map. DETAILED DESCRIPTION OF THE INVENTION

[0026] The data processing method and data processing system of the present invention will be described below with reference to the drawings. In each embodiment described below, the same reference numerals are used to designate the same parts as those in the previous drawings, and detailed descriptions thereof will be omitted. Differences will be mainly described.

[0027] The data processing system according to the present invention is comprised of multiple functional modules and can be installed as software in an independent device such as a computer having a CPU (central processing unit) and memory, or can be installed in a distributed manner in multiple devices, and is realized by a processor executing each functional module of the data processing system stored in the memory. The circuitry realizing the data processing system can send, receive, or collect data via a network such as the Internet.

[0028] In the following description, the data processing method and data processing system will be described as being used for detecting a living organism inside an automobile. However, the scope of application of the data processing method and data processing system of the present invention is not limited to this. For example, the data processing method and data processing system may be used for detecting a living organism inside a means of transportation other than an automobile. In addition, in the description of the present invention, the term "living organism" does not mean a human being in the narrow sense, but also includes various living organisms such as pets.

[0029] (Data Processing System 1) The data processing system 1 will be described below with reference to Fig. 1. Fig. 1 is a block diagram showing an example of the configuration of the data processing system 1.

[0030] As shown in FIG. 1, the data processing system 1 includes a communication IF 11, a data processing unit 12, and a memory 13.

[0031] The communication IF 11 is an interface used for communication of various data or information. The data processing system 1 is communicably connected to, for example, a millimeter wave radar 2 and an operation terminal 3 external to the data processing system 1 via the communication IF 11. Furthermore, communication of various data or information is performed between the components included in the data processing system 1 via the communication IF 11.

[0032] The data processing unit 12 is a functional unit for processing received data. The data processing unit 12 processes the data collected using the millimeter wave radar 2 by a data processing method to be described later.

[0033] The memory 13 is a device for storing various data or information. The memory 13 may be a processor-readable storage medium (e.g., a magnetic storage medium, an electromagnetic storage medium, an optical storage medium, or a semiconductor memory), or may be a drive device for reading and writing data or information from and to the storage medium. The memory 13 stores, for example, a computer program required for causing the data processing unit 12 to implement the data processing method described below.

[0034] In the above description, an example has been given in which the data processing system 1, the millimeter-wave radar 2, and the operation terminal 3 are each configured independently. However, this is not limiting, and for example, the data processing system 1 may be mounted on the millimeter-wave radar 2 or the operation terminal 3. It is sufficient that the data processing system 1 can process the data collected using the millimeter-wave radar 2 using a data processing method described below.

[0035] The millimeter-wave radar 2 will be described with reference to Figures 2A and 2B. Figure 2A is a block diagram showing an example of the configuration of the millimeter-wave radar 2. Figure 2B is a schematic diagram showing the distribution of the spatial range of the millimeter-wave radar within the cabin of a vehicle.

[0036] The millimeter-wave radar 2 is a radar device that transmits millimeter-wave radar signals and analyzes collected data (radar signals) to detect information such as the presence, location, and time of a living body inside a vehicle. As shown in Fig. 2A, the millimeter-wave radar 2 includes a radar antenna 21, a radar chip (ASIC) 22, a power supply 23, and a communication IF 24.

[0037] As shown in Fig. 2B, the spatial range of the millimeter-wave radar 2 is set, for example, around the cabin (for example, the dashed circle in Fig. 2B). This is because the effective detection range of the millimeter-wave radar 2 is usually within 2 meters and there are usually many obstacles inside the cabin. Furthermore, the set position and number of spatial ranges of the millimeter-wave radar 2 are not limited to this and may be other set positions and numbers.

[0038] The operation terminal 3 is a terminal operated by a user (e.g., a driver). The operation terminal 3 can also function as a client of the data processing system 1. The operation terminal 3 may be a smartphone, a tablet terminal, an in-vehicle terminal, a wearable terminal, a mobile terminal, a handheld terminal, or the like. The operation terminal 3 also has an input interface that accepts various operations from the user. The operation terminal 3 may also have a display device that displays various data or information. The display device may be a touch panel display device that also functions as an input interface.

[0039] Specifically, the operation terminal 3 accepts the user's confirmation result that no living organism is present in the vehicle, and transmits the user's confirmation result that no living organism is present in the vehicle to the data processing system 1. When it is necessary to collect in-vehicle environment data (described later), the operation terminal 3 can notify the user by a display device or the like, thereby allowing the user to confirm that no living organism is present in the vehicle. The operation terminal 3 periodically notifies the user that it is necessary to collect in-vehicle environment data, for example, based on a preset time interval. Furthermore, when the millimeter-wave radar 2 detects the presence of a living organism in the vehicle, the operation terminal 3 can also issue an alarm to the user or another terminal by a display device or the like, informing that a living organism is present in the vehicle.

[0040] (The process of detecting a living body inside the vehicle) The process of detecting a living body inside a vehicle will now be described with reference to Fig. 3. Fig. 3 is a flow chart showing the process of detecting a living body inside a vehicle.

[0041] In step S1, the data processing system 1 determines whether the door is closed. If it determines that the door is closed ("YES" in step S1), the process proceeds to step S2. If it determines that the door is not closed ("NO" in step S1), the process continues with the determination in step S1. Also, in step S1, the data processing system may determine whether the door is closed and locked instead of determining whether the door is closed or locked, and proceed to step S2 only if the door is closed and locked, and otherwise continue with the determination in step S1.

[0042] In step S2, the millimeter-wave radar 2 collects in-vehicle actual measurement data. Here, "in-vehicle actual measurement data" refers to data collected by the millimeter-wave radar 2 to detect whether or not a living body is present inside the vehicle, for example, every time the driver leaves the vehicle and closes the door. Then, the process proceeds to step S3.

[0043] In step S3, the data processing system 1 processes the in-vehicle measured data collected by the millimeter-wave radar 2 in step S2 based on the noise map, and removes noise from the in-vehicle measured data. Here, the noise map used by the data processing system 1 is generated by a noise map generation process described later. The noise map generation process and the method for removing noise from the in-vehicle measured data based on the noise map will be described in detail later. Next, the process proceeds to step S4.

[0044] In step S4, the millimeter wave radar 2 performs data filtering on the in-vehicle measured data after noise has been removed by the data processing system 1 in step S3. Then, the process proceeds to step S5.

[0045] In step S5, the millimeter wave radar 2 selects data from living bodies from the in-vehicle measured data after data filtering in step S4. Then, the process proceeds to step S6.

[0046] In step S6, the millimeter wave radar 2 performs a scoring process on the data from the living body selected in step S5, and detects whether or not a living body is present inside the vehicle based on the score. Then, the process proceeds to step S7.

[0047] In step S7, if the millimeter wave radar 2 detects that a living body is present inside the vehicle ("YES" in step S7), the process proceeds to step S8. If the millimeter wave radar 2 detects that a living body is not present inside the vehicle ("NO" in step S7), the process ends.

[0048] In step S8, the millimeter wave radar 2 issues an alarm to the user or another terminal via, for example, a display device of the operation terminal 3, indicating that a living body is present inside the vehicle. Then, the process ends.

[0049] Here, in the present invention, there are no limitations on the methods of "data filtering," "selection of data from living organisms," "scoring processing," etc. mentioned in the above description, and any known method can be used.

[0050] For example, when step S4 is performed, data filtering can be performed by amplifying the characteristic parts of the in-vehicle measured data using a filtering algorithm.

[0051] When performing step S6, the data from the living body selected in step S5 is scored exist_score (0 to 1.0), and a threshold value thresh_normal (e.g., 0.6) is preset. If the score exist_score is greater than the threshold value thresh_normal, it is determined that a living body is present in the vehicle. If the score exist_score is equal to or less than the threshold value thresh_normal, it is determined that a living body is not present in the vehicle. Normally, under ideal conditions, when a driver is present in the vehicle, the score exist_score reaches 0.8. In contrast, in a particular scene or time with strong noise, the score exist_score drops to 0.7 or less.

[0052] (Noise map generation process) The noise map generation process will be described below with reference to Fig. 4. Fig. 4 is a flowchart showing the noise map generation process. In step S3 shown in Fig. 3, the data processing system 1 removes noise from the in-vehicle measurement data based on the noise map generated by the noise map generation process (steps S301 to S306) shown in Fig. 4.

[0053] In the present invention, the "in-vehicle environment data" refers to data indicating the environment inside the vehicle collected by the millimeter-wave radar 2 when it is confirmed that no living body is present inside the vehicle based on the user's operation on the operation terminal 3. The "in-vehicle environment data" can be considered to reflect a collection of noise data inside the vehicle.

[0054] The data processing system 1 periodically notifies the user via the operation terminal 3 whether or not a noise map needs to be generated, and collects in-vehicle environment data when the user selects "necessary." The period for periodically notifying the user via the operation terminal 3 whether or not a noise map needs to be generated may be, for example, once a month. Furthermore, the data processing system 1 may generate a new noise map when the time elapsed since the previous noise map generation exceeds a preset time interval.

[0055] In step S301, the data processing system 1 determines whether the user has confirmed that no living organism is present inside the vehicle. Here, the data processing system 1 notifies the user via, for example, the operation terminal 3, so that the user can confirm that no living organism is present inside the vehicle. If it is confirmed that no living organism is present inside the vehicle based on the user's operation on the operation terminal 3, the process proceeds to step S302.

[0056] In step S302, the millimeter-wave radar 2 collects in-vehicle environment data. At this time, since the user confirmed in step S301 that no living organism is present in the vehicle, the in-vehicle environment data collected by the millimeter-wave radar 2 does not include information from living organisms, but includes information from noise in the in-vehicle environment. In addition, the in-vehicle environment data collected by the millimeter-wave radar 2 includes at least one of time domain data, frequency domain data, and spatial data. Next, the process proceeds to steps S303A, S304A, and S305A.

[0057] In step S303A, the data processing system 1 generates time domain data from the in-vehicle environment data collected by the millimeter-wave radar 2. Here, the generated time domain data includes data indicating time changes in the in-vehicle environment data. In addition, the data processing system 1 generates the time domain data by, for example, analog-to-digital conversion or signal queueing. Then, the process proceeds to step S303B.

[0058] In step S303B, the data processing system 1 identifies noise information from the time-domain data generated in step S303A. The method for "identifying noise" is not limited in any way, and any known method can be used. The "noise information" in the present invention includes, for example, information indicating the time, type, source, and location of noise. Then, the process proceeds to step S303C.

[0059] In step S303C, the data processing system 1 identifies the time range of noise occurrence based on the noise information identified in step S303B.

[0060] Specifically, when noise information is identified from time domain data, the noise information is, for example, information indicating a change in noise in the time domain.

[0061] For example, when a door is closed, the vehicle generates large vibrations and noise within a short period of time. The collected time-domain data of the in-vehicle environment data is characterized by a large amplitude waveform, and the larger the amplitude, the more likely it is that such noise is present. By identifying waveforms with large amplitude values ​​within a short period of time, it is possible to identify time-domain information about noise caused by closing the door from the time-domain data. The time range of noise occurrence can then be identified based on changes in the time-domain data. For example, a section with large amplitude values ​​of the waveform can be identified as the time range of noise occurrence. Furthermore, it is possible to distinguish between permanent noise and transient noise in the time-domain data of the in-vehicle environment data. This is because transient noise causes large amplitude changes in the waveform of the time-domain data within a short period of time. Next, the process proceeds to step S306.

[0062] In step S304A, the data processing system 1 generates frequency domain data of the in-vehicle environment data collected by the millimeter-wave radar 2. Here, the generated frequency domain data includes data indicating frequency changes of the in-vehicle environment data. In addition, the data processing system 1 generates the frequency domain data by, for example, a filtering and integration method. Then, the process proceeds to step S304B.

[0063] In step S304B, the data processing system 1 identifies noise information from the frequency domain data generated in step S304A, and then proceeds to step S304C.

[0064] In step S304C, the data processing system 1 identifies the frequency range of noise occurrence based on the noise information identified in step S304B.

[0065] Specifically, when noise information is identified from frequency domain data, the noise information is, for example, information indicating changes in the frequency domain of noise.

[0066] For example, when liquid (e.g., mineral water) is placed inside a vehicle, the liquid continuously generates noise of a specific frequency similar to the breathing of a living organism. The collected frequency domain data of the interior environment data is characterized by a high SNR (signal-to-noise ratio) at a specific frequency, and the higher the SNR, the more likely such noise is present. By identifying the specific frequency at which the high SNR occurs, it is possible to identify, from the frequency domain data, information on the frequency domain of noise caused by the liquid placed inside the vehicle. Furthermore, for example, when a decorative part is suspended inside the vehicle, the swinging of the decorative part also generates noise of a specific frequency. In this case, the frequency range in which the fixed vibration noise occurs can be identified based on changes in the frequency domain data. For example, the specific frequency or a specific frequency range at which the high SNR occurs can be identified as the frequency range in which the noise occurs. Next, the process proceeds to step S306.

[0067] In step S305A, the data processing system 1 generates spatial data from the in-vehicle environment data collected by the millimeter-wave radar 2. Here, the generated spatial data includes data indicating spatial changes in the in-vehicle environment data. In addition, the data processing system 1 calculates the spatial distance using, for example, a timer and a signal strength matrix offset amount, and uses the calculated spatial data. Then, the process proceeds to step S305B.

[0068] In step S305B, the data processing system 1 identifies noise information from the spatial data generated in step S305A, and then proceeds to step S305C.

[0069] In step S305C, the data processing system 1 identifies the location range of the noise occurrence based on the noise information identified in step S305B.

[0070] Specifically, when noise information is identified from spatial data, the noise information is, for example, information indicating a change in signal intensity in a noise space.

[0071] For example, the vehicle structure and various objects placed on it can cause a multipath effect, and at certain spatial locations, the signal strength of a radar signal reflected by the vehicle structure or objects can potentially reach the signal strength of a radar signal reflected by a living body. In addition to the localized high signal strength caused by reflections from the vehicle structure and objects, diffraction or refraction can also cause localized high signal strength. Any of these localized high signal strengths can reach the signal strength of a radar signal reflected by a living body. The spatial data of the collected in-vehicle environment data has characteristics that generate localized high signal strengths (amplitude values), and the higher the signal strength, the more likely such noise is to be present. By identifying the locally occurring high signal strengths, spatial information about noise caused by the vehicle structure and various objects placed on it can be identified from the spatial data. Next, proceed to step S306.

[0072] The process of removing noise from the in-vehicle measured data based on the noise map will be described with reference to Fig. 5. Fig. 5 is a schematic diagram showing the process of removing noise from the in-vehicle measured data based on the noise map.

[0073] In step S306, the data processing system 1 generates a noise map based on the in-vehicle environment data collected by the millimeter-wave radar 2 when the user confirms that no living body is present in the vehicle. Specifically, the noise map is generated based on the time-domain data, frequency-domain data, and spatial data of the in-vehicle environment data. Here, the "noise map" can be considered a database containing noise information. The noise map stores noise information and noise reduction methods for removing the noise in association with each other. As shown in FIG. 5 , the noise map is generated based on the time-domain data, frequency-domain data, and spatial data of the in-vehicle environment data. Therefore, the noise information in the noise map can be divided into noise time-domain features, noise frequency-domain features, and noise spatial features. The noise time-domain features include the noise identified from the time-domain data and the time range of the noise. The noise frequency-domain features include the noise identified from the frequency-domain data and the frequency range of the noise. The noise spatial features include the noise identified from the spatial data and the location range of the noise. Then, the process ends.

[0074] As a result, the data processing system 1 generates a noise map based on the in-vehicle environment data collected when the user confirms that no living body is present in the vehicle. In addition, in step S3 of the process of detecting a living body in the vehicle, the data processing system 1 removes noise from the in-vehicle actual measurement data based on the generated noise map.

[0075] Specifically, when removing noise from the in-vehicle measured data, the data processing system 1 divides the in-vehicle measured data in three dimensions: time domain, frequency domain, and space. For example, as shown in Fig. 5, the data processing system 1 divides the in-vehicle measured data into Range bin 1, Range bin 2, Range bin 3, and Range bin 4 according to Range bins (spatial ranges) in the spatial dimension. In this case, the right diagram in Fig. 5 shows the time domain data and frequency domain data of the in-vehicle measured data corresponding to the divided Range bins 1 to 4, respectively. Furthermore, the division method and division results of the in-vehicle measured data are not limited to this. For example, the in-vehicle measured data may be divided into Range bin 1 to Range bin 8.

[0076] Next, the data processing system 1 removes noise from the in-vehicle measured data based on the noise information in the noise map (noise time domain features, noise frequency domain features, and noise spatial features) and the noise reduction method associated therewith.

[0077] For example, as can be seen from the noise time domain characteristics in the noise map, the range indicated by the dashed line frame in the time domain dimension is the time range of noise occurrence. In this case, the data processing system 1 removes data within this time range from the in-vehicle actual measurement data based on information indicating the time range of noise occurrence in the time domain data of the in-vehicle environment data included in the noise map. This results in the time domain dimension data of the processed in-vehicle actual measurement data shown in FIG. 5.

[0078] As can be seen from the noise frequency domain characteristics in the noise map, the range indicated by the dashed-dotted line frame in the frequency domain dimension is the frequency range of noise occurrence. At this time, the data processing system 1 removes data in this frequency range from the in-vehicle actual measurement data based on information indicating the frequency range of noise occurrence in the frequency domain data of the in-vehicle environment data included in the noise map. This results in the frequency domain dimension data of the processed in-vehicle actual measurement data shown in FIG. 5.

[0079] As can be seen from the noise spatial characteristics in the noise map, in the spatial dimension, Range bin 3 shown in a gray frame is the positional range of noise occurrence. At this time, the data processing system 1 removes data in this positional range from the in-vehicle actually measured data based on information indicating the positional range of noise occurrence in the spatial data of the in-vehicle environment data included in the noise map. This results in the spatial dimension data of the processed in-vehicle actually measured data shown in FIG. 5.

[0080] As a result, the data processing system 1 removes noise from the in-vehicle measured data based on the noise map. Then, the process in which the data processing system 1 processes the in-vehicle measured data to remove noise from the in-vehicle measured data ends, and the process proceeds to step S4.

[0081] (Effects of the Invention) The data processing method and system according to the present invention can improve the accuracy of detecting a living body in a vehicle, accurately remove noise, reduce the amount of calculation required for detecting a living body in a vehicle, and improve processing efficiency.

[0082] Specifically, when detecting whether a living organism is present inside the vehicle based on the in-vehicle measured data, the data processing system 1 removes at least one of the time range, frequency range, and position range of noise occurrence in the in-vehicle measured data, thereby improving the accuracy of detecting a living organism inside the vehicle based on the in-vehicle measured data. Furthermore, because noise data is removed from the in-vehicle measured data, the amount of calculation required to detect a living organism inside the vehicle based on the in-vehicle measured data can be reduced, improving processing efficiency.

[0083] Furthermore, in the noise map generated based on the in-vehicle environment data, noise information and noise reduction methods are stored in correspondence with each other, so that the corresponding noise reduction method can be appropriately used depending on various noise situations, and the suitability of the noise reduction method is high.

[0084] Furthermore, when using conventional vehicle interior detection methods, the score of the detection result approaches a preset threshold due to the above-mentioned cases of closing the door, placing liquid inside the vehicle, or having decorative parts hanging inside the vehicle. However, in the present invention, noise data in the vehicle interior measurement data is removed based on the noise map, thereby improving the accuracy of vehicle interior detection.

[0085] Furthermore, because the data processing method of the present invention is performed when it is determined that the door is closed, it is possible to fully utilize the computational resources of the millimeter-wave radar 2 when the vehicle is idling. Conventional methods for detecting living beings inside a vehicle require the use of high-load algorithms such as FFT, which causes the computational resources of the millimeter-wave radar 2 to become almost saturated. However, the data processing method of the present invention removes noise data from the actual measurement data inside the vehicle based on a noise map, thereby optimizing computational resources, reducing processing time, and improving processing efficiency.

[0086] Furthermore, in conventional data processing methods, noise is removed from the actual vehicle interior measurement data using preset vehicle interior environment data at the time of shipment each time a living organism is detected inside the vehicle. However, as described above, the vehicle interior environment changes when the door is closed, liquid is placed inside the vehicle, decorative parts are hung inside the vehicle, etc. In such cases, using the vehicle interior environment data at the time of shipment cannot accurately remove noise, resulting in a decrease in detection accuracy. In contrast, the data processing method of the present invention collects vehicle interior environment data when it is confirmed that no living organism is present inside the vehicle, generates a noise map, and removes noise from the actual vehicle interior measurement data based on the noise map, thereby enabling accurate noise removal.

[0087] The above embodiments are merely examples and are not intended to limit the present invention. Those skilled in the art may appropriately add, delete, or modify components of the above embodiments, but any such modifications are within the scope of the present invention as long as they incorporate the technical concept of the present invention.

[0088] For example, the generated noise map may be uploaded to a cloud server and trained by a learning model deployed on the cloud server, and the trained noise map may be distributed to each vehicle or terminal. [Explanation of symbols]

[0089] 1. Data Processing System 11 Communication Interface 12 Data processing section 13. Memory 2. Millimeter wave radar 21 Radar Antenna 22 radar chip 23 Power supply 24 Communication Interface 3 Operation terminal

Claims

1. A data processing method for processing data collected using a millimeter wave radar, comprising: collecting in-vehicle actual measurement data; and removing noise from the in-vehicle actual measurement data based on a noise map; The data processing method is characterized in that the noise map is generated based on in-vehicle environment data collected when it is confirmed that no living body is present in the vehicle.

2. 2. The data processing method according to claim 1, wherein the in-vehicle environment data includes at least one of time domain data, frequency domain data, and spatial data.

3. In the step of removing noise from the in-vehicle actual measurement data, identifying noise information in the in-vehicle actual measurement data based on the time domain data, the frequency domain data, and the spatial data, based on the noise map; 3. The data processing method according to claim 2, wherein the noise is removed using a noise reduction method corresponding to the noise information.

4. In the step of removing noise from the in-vehicle actual measurement data, 4. The data processing method according to claim 3, wherein data within the time range of the in-vehicle measured data is removed based on information indicating a time range of the occurrence of the noise in the time domain data.

5. In the step of removing noise from the in-vehicle actual measurement data, 4. The data processing method according to claim 3, wherein data in the frequency range of the noise in the in-vehicle measured data is removed based on information indicating the frequency range of the noise in the frequency domain data.

6. In the step of removing noise from the in-vehicle actual measurement data, 4. The data processing method according to claim 3, wherein data within the position range of the noise occurrence in the spatial data is removed based on information indicating the position range of the noise occurrence in the spatial data.

7. In the step of confirming that no living body is present in the vehicle, 2. The data processing method according to claim 1, wherein the absence of the living body in the vehicle is confirmed based on an operation by a user.

8. 8. The data processing method according to claim 1, wherein a characteristic portion of the in-vehicle measured data is amplified by a filtering algorithm.

9. A data processing system for processing data collected using millimeter wave radar, A data processing system using the data processing method according to any one of claims 1 to 7.

Images