Multi-label spatial fusion identity recognition method, system, medium and terminal

Abstract

The present invention provides a multi-tag spatial fusion identity recognition method, system, medium and terminal, which obtains original phase data through an RFID tag array and a reader / writer deployed in a traffic area; pre-processes the original phase data and extracts walking-related data of an object to be identified; obtains sample data of a sample object, and evaluates the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain the distinguishing ability value and anti-interference ability value of each tag; based on the distinguishing ability value and anti-interference ability value, a multi-dimensional weighted dynamic time warping algorithm is used to evaluate the similarity between the walking-related data of the object to be identified and the sample data, so as to obtain the identity information of the object to be identified. The present invention realizes accurate identification of a person's identity when gait factors are inconsistent, maximizes the advantages of the spatial diversity of the tag array, and has good recognition accuracy and robustness.

Description

Technical Field

[0001] The present invention relates to the field of intelligent identity recognition, and in particular to an identity recognition method, system, medium and terminal for multi-label space fusion. Background Art

[0002] In recent years, the human behavior sensing technology based on wireless signals has attracted widespread attention from the academic community for its non-invasive, easy deployment and privacy protection characteristics. A large number of innovative applications have sprung up, such as fall detection, gesture recognition, and even fine-grained lip recognition, vital sign monitoring, etc. However, if the identity information of the person performing the action cannot be obtained, these applications will greatly reduce the experience of the object, and even fail to work in some scenarios. For example, if the motion recognition system cannot identify the identity of the person entering the room, it will not be able to further provide personalized services such as TV program recommendations or room lighting temperature adjustment, and it will not be able to conduct personalized analysis of their behavior. Therefore, identity recognition is an important prerequisite for wireless behavior sensing technology in smart spaces.

[0003] Compared with other biometric technologies (face recognition, fingerprint recognition, etc.), gait recognition has the advantages of long-distance observation and no need for cooperation. Therefore, existing wireless identity recognition systems generally identify people based on their gait characteristics when they walk. The basic principle is that when people walk near wireless devices, due to differences in gait, body shape, etc., different people will have very different effects on wireless signals. Therefore, the fluctuation characteristics of these wireless signals contain information that can represent people's identity. However, in actual scenarios, people's gait factors are rich in diversity, such as backpacks, making phone calls, carrying laptops, or changes in appearance and clothing. Inconsistency in gait factors will cause changes in wireless signal fluctuations when people walk, which will seriously affect recognition performance. Existing studies all assume that the gait factors in the training phase and the test phase are completely consistent, but this assumption is obviously not true in actual scenarios. If all possible gait factors are trained in the training phase, it will cost a lot of manpower and lack scalability, which is not advisable in practice. At present, gait recognition research based on wireless signals uses the channel state information (CSI) of WiFi signals to characterize human gait characteristics. Although the multi-carrier characteristics of WiFi can provide rich frequency diversity, the fusion of multi-frequency CSI information still cannot effectively eliminate the interference of gait factors. This is mainly because multiple subcarriers of WiFi signals are transmitted through the same propagation path. When the gait factor changes and the propagation path changes, the CSI information of all frequencies will be affected. Summary of the invention

[0004] In view of the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a multi-label spatial fusion identity recognition method, system, medium and terminal, which are used to solve the problem of interference of multiple gait factors on identity recognition in the prior art.

[0005] To achieve the above-mentioned purpose and other related purposes, the first aspect of the present invention provides an identity recognition method with multi-tag spatial fusion, including: obtaining original phase data through an RFID tag array and a reader / writer deployed in a traffic area; preprocessing the original phase data and extracting walking-related data of the object to be identified; obtaining sample data of a sample object, and evaluating the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain the distinguishing ability value and anti-interference ability value of each tag; based on the distinguishing ability value and anti-interference ability value, using a multi-dimensional weighted dynamic time warping algorithm to evaluate the similarity between the walking-related data of the object to be identified and the sample data, so as to obtain the identity information of the object to be identified.

[0006] In some embodiments of the first aspect of the present invention, the multidimensional weighted dynamic time warping algorithm includes: based on the sample data and the walking-related data, obtaining a sequence alignment distance matrix that satisfies a preset window restriction condition; the window restriction condition is: the difference between the row number and the column number of each element in the matrix does not exceed a preset value; based on the sequence alignment matrix, combined with the distinguishing ability value and anti-interference ability value of each label, obtaining the multidimensional weighted dynamic time warping distance between the object to be identified and the sample; selecting the smallest multiple distances from the multidimensional weighted dynamic time warping distances and performing weighted calculation to obtain the similarity.

[0007] In some embodiments of the first aspect of the present invention, the distinguishing ability value includes: a global distinguishing ability value and a local distinguishing ability value; the method also includes: calculating the similarity based on the global distinguishing ability value, and extracting multiple groups of high-similar sample data with the highest similarity to the walking-related data; based on the local distinguishing ability value, respectively calculating the similarity between the walking-related data and multiple groups of the high-similar sample data to obtain the most similar sample data; the sample object identity information corresponding to the most similar sample data is the identity information of the object to be identified.

[0008] In some embodiments of the first aspect of the present invention, the walking-related data includes walking direction data; the acquisition method thereof includes: collecting the directional phase signal of a directional tag deployed on one side of the tag array; based on the directional phase signal, acquiring the time for the object to be identified to pass through the line-of-sight propagation path of the tag array and the directional tag to obtain the walking direction data of the object to be identified.

[0009] In some embodiments of the first aspect of the present invention, the method for acquiring the walking-related data includes: based on the line-of-sight propagation path of the tag array and the reader / writer, performing feature extraction on the pre-processed raw phase data to acquire the walking-related data.

[0010] In some embodiments of the first aspect of the present invention, the method for obtaining the distinguishing ability evaluation value includes: calculating the dynamic time warping distance of each label based on the sample data of the sample object, including internal distance and external distance; normalizing the internal distance and external distance to obtain the global distinguishing ability value of each label for all sample objects; and obtaining the local distinguishing ability value of each label for every two sample objects based on the internal distance and external distance.

[0011] In some embodiments of the first aspect of the present invention, the method for acquiring the walking-related data includes: detecting whether an action occurs near the tag array based on the relative entropy of the original phase data, and extracting corresponding action data; judging whether the action occurs is a walking action through the passage area based on the time-frequency characteristics of the action data, and extracting corresponding walking-related data.

[0012] To achieve the above-mentioned purpose and other related purposes, the second aspect of the present invention provides a multi-tag spatial fusion identity recognition system, including: an original data acquisition module, which acquires original phase data through an RFID tag array and a reader / writer deployed in a traffic area; a walking data extraction module, which is used to preprocess the original phase data and extract walking-related data of the object to be identified; a tag evaluation module, which is used to acquire sample data of a sample object, and evaluate the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain the distinguishing ability value and anti-interference ability value of each tag; an identity recognition module, which uses a multi-dimensional weighted dynamic time warping algorithm to evaluate the similarity between the walking-related data of the object to be identified and the sample data based on the distinguishing ability value and anti-interference ability value, so as to obtain the identity information of the object to be identified.

[0013] To achieve the above-mentioned purpose and other related purposes, the third aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the identity recognition method of multi-label spatial fusion is implemented.

[0014] To achieve the above-mentioned purpose and other related purposes, the fourth aspect of the present invention provides an electronic terminal, including: a processor and a memory; the memory is used to store computer programs, and the processor is used to execute the computer programs stored in the memory, so that the terminal executes the multi-label spatial fusion identity recognition method.

[0015] As described above, the multi-tag spatial fusion identity recognition method, system, medium and terminal involved in the present invention have the following beneficial effects: using RFID equipment to realize non-invasive identity recognition suitable for inconsistent gait factors; in some scenarios where straight walking is required, such as doors, corridors, etc., the RFID reader antenna and tag array are deployed on both sides of the walking route. When someone passes by, the phase information of the tag return signal can well reflect the individual differences of people; moreover, the spatial diversity of the tag array can provide rich spatiotemporal information for increasing the differences in signal characteristics of different objects, while reducing the interference of gait factors, with extremely high identity recognition accuracy, and the system has good robustness. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 Shown is a flow chart of an identity recognition method of multi-label spatial fusion in one embodiment of the present invention.

[0017] Figure 2 Shown is a flow chart of another identity recognition method of multi-label spatial fusion in one embodiment of the present invention.

[0018] Figure 3 Shown is a structural schematic diagram of an identity recognition system with multi-label spatial fusion in one embodiment of the present invention.

[0019] Figure 4 Shown is a schematic structural diagram of an electronic terminal in one embodiment of the present invention. DETAILED DESCRIPTION

[0020] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the following embodiments and features in the embodiments can be combined with each other without conflict.

[0021] It should be noted that in the following description, reference is made to the accompanying drawings, which describe several embodiments of the present invention. It should be understood that other embodiments may also be used, and that mechanical composition, structural, electrical, and operational changes may be made without departing from the spirit and scope of the present invention. The following detailed description should not be considered restrictive, and the scope of the embodiments of the present invention is limited only by the claims of the published patents. The terms used herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0022] Furthermore, as used in this article, the singular forms "a", "an" and "the" are intended to include the plural forms as well, unless there is an indication to the contrary in the context. It should be further understood that the terms "comprise", "include" indicate the presence of the described features, operations, elements, components, items, kinds, and / or groups, but do not exclude the presence, occurrence or addition of one or more other features, operations, elements, components, items, kinds, and / or groups. The terms "or" and "and / or" used herein are interpreted as inclusive, or mean any one or any combination. Therefore, "A, B or C" or "A, B and / or C" means "any of the following: A; B; C; A and B; A and C; B and C; A, B and C". Exceptions to this definition will only occur when the combination of elements, functions or operations is inherently mutually exclusive in some way.

[0023] The present invention proposes a multi-label space fusion identity recognition method, device, system, medium and terminal, which are used to solve the problem of interference of multiple gait factors on identity recognition in the prior art.

[0024] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the technical solution in the embodiments of the present invention is further described in detail through the following embodiments and in conjunction with the accompanying drawings. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] Embodiment 1

[0026] like Figure 1 As shown, an embodiment of the present invention proposes a flow chart of an identity recognition method of multi-label space fusion, which includes:

[0027] Step S11. Obtain the original phase data through the RFID tag array and reader / writer deployed in the passage area. The passage area can be various doors, entrances, corridors and other areas. The RFID tag array is a two-dimensional planar array composed of RFID tags. The number of tags, number of rows and columns, arrangement direction, tag spacing, etc. need to be flexibly adjusted according to the actual application scenario to collect sufficient data for effective identification of the passing objects. The antenna of the reader / writer is arranged corresponding to the RFID tag array, and its beam width covers all tags in the tag array. When the object to be identified passes through the passage area, the reader / writer reads the phase change information that can characterize individual differences from the tag array, which can then be used to identify the passing object.

[0028] Step S12: Preprocess the raw phase data and extract the walking-related data of the object to be identified. The raw phase data collected from the reader can first be preprocessed by phase unwrapping, smoothing, etc., and then the phase data related to human walking can be extracted.

[0029] In some examples, the preprocessing process of the raw phase data can be described as follows: After the system receives the raw phase data of the tag from the reader, the first step is to unwrap it. This is because the phase data received by the reader will only change within the range of 0 to 2π. Due to the high reading rate of the reader for the tag (for 5 tags, the reading rate can reach nearly 70 times per second for each tag), the frequency of tag phase changes caused by human body movement is much lower than the reading rate of the reader. An example is given below to prove this: when a person with an RFID tag is facing the reader antenna and is approaching or moving away from it, if the reader reads the tag 50 times per second, the interval between two consecutive readings is 20ms. If the person's movement speed is 1m / s, the moving distance between two consecutive readings of the tag is 0.02m, and the phase change is about 0.8rad. In fact, since the impact of people on the tag is caused by reflection and diffraction, the phase change will be smaller. Therefore, it can be determined that the difference in phase data of the same tag twice in a row does not exceed π. When the change exceeds π, it is considered that the phase has jumped, and an operation of adding 2π or subtracting 2π is performed to keep the change within the interval [-π,π].

[0030] Furthermore, after phase unwrapping, the data is initially smoothed and denoised using a Hamel outlier filter and a median filter; the DC component in the phase waveform is filtered out by calculating the phase data in a stationary state; and the phase data is processed using a linear interpolation method (for example, the data is interpolated every 10 seconds). Since the reader uses a dynamic time-slot Aloha algorithm to read the tag, the tag's phase data is non-uniformly distributed in the time domain, and linear interpolation processing is beneficial to improving the efficiency of subsequent data analysis and processing.

[0031] In a preferred implementation of this embodiment, based on the relative entropy (KL divergence, Kullback-Leibler divergence) of the original phase data, it is detected whether an action occurs near the tag array, and the corresponding action data is extracted; based on the time-frequency characteristics of the action data, it is determined whether the action that occurs is a walking action through the passage area, and the corresponding walking-related data is extracted.

[0032] Among them, detecting whether there is action near the tag array (such as whether there is human movement) based on KL divergence includes: first dividing the phase waveform into phase frames of preset duration and overlap (such as duration 1s, 50% overlap), and then calculating the KL divergence value of two consecutive frames; extracting the motion-related data by comparing the calculated KL divergence value with the preset threshold. KL divergence is a measure that describes the difference between two probability distributions. The closer the distribution is, the closer the KL divergence value is to 0, and vice versa. Therefore, when no one is moving, the KL divergence value will basically remain near 0, and when someone is moving, it will become a larger value; by comparing the size of the KL divergence value, the data related to human movement can be extracted. Since KL divergence can only detect the occurrence of movement and cannot guarantee that all data related to movement can be extracted, in order to prevent data loss, it is preferred to set a looser threshold to include more data. Moreover, since the degree of change of the tag is strongly related to the "distance from the reader to the antenna", the preset threshold of the KL divergence is negatively correlated with the distance between the reader and the antenna in the actual application scenario, that is, the farther the distance between the reader and the antenna, the smaller the preset threshold of the KL divergence.

[0033] Furthermore, it is determined whether the action that occurs is a walking action through the passage area, which includes: using some lightweight time-frequency domain statistical features for judgment, these features include time domain features: maximum value, minimum value, average value, skewness, kurtosis, variance, zero-crossing rate, frequency domain features: spectral entropy, spectral energy and frequency value with maximum energy; based on the time-frequency statistical feature values ​​and using a decision tree algorithm to determine whether a walking action occurs.

[0034] It should be noted that the conventional application scenarios of tag arrays and reader antennas are corridors, doors and other areas. Although these areas are relatively small, people's activities around them will still cause signal fluctuations, and the movement of tags away from the other side of the reader antenna will also cause obvious signal fluctuations. In a home environment, many activities may also occur around the device, such as sitting down, standing up, walking around, or not completely passing through the reader antenna and tag array, but just walking halfway and then walking back. Therefore, it is necessary to identify the movement before identity recognition to determine whether it is a walking behavior that completely passes through the reader antenna and tag array. Since the signal differences caused by different movements are relatively obvious, preliminary screening of relevant data by whether there is any action around the tag array and whether it is a walking action can not only avoid data redundancy, but also help improve the accuracy and efficiency of identity recognition.

[0035] In a preferred implementation of this embodiment, based on the line of sight (LOS) path of the tag array and the reader, feature extraction is performed on the pre-processed raw phase data to obtain the walking-related data. The theoretical basis is as follows: when the center position of a person passes through the line of sight path of the reader and the tag array, the phase of the tag return signal is the largest. This implementation can achieve more fine-grained walking-related data extraction, which is beneficial to saving computing resources and accelerating the rate of related data processing.

[0036] In some examples, the data extraction process based on the LOS path can be specifically operated as follows: divide the acquired phase data into frames of preset duration and overlap (such as 0.1s duration and 50% overlap), and find the frame with the largest energy; take the center moment of the frame as the time when the center position of the object to be identified passes through the LOS path, and then extract the phase data in a certain interval around it.

[0037] In a preferred implementation manner of this embodiment, the walking-related data includes walking direction data; the acquisition method thereof includes: collecting the directional phase signal of a directional tag deployed on one side of the tag array; based on the directional phase signal, acquiring the time for the object to be identified to pass through the line-of-sight propagation path of the tag array and the directional tag to obtain the walking direction data of the object to be identified.

[0038] In some application scenarios of the present invention, there are only two walking directions. If the walking directions of a person are opposite, the reflection and diffraction paths of the tag signal will be very different, so that the phase waveform will also show obvious differences. Therefore, the data of the object to be identified and the sample data of the two directions in the training set can be matched to determine the walking direction while performing identity recognition. On the other hand, if the walking direction of a person can be determined before identity recognition, the amount of calculation and error rate during matching will be greatly reduced.

[0039] Specifically, by utilizing the easy deployment feature of RFID tags, one or more additional tags are deployed on one side (left or right) of the tag array as direction tags, and the walking direction can be determined by the LOS time when a person passes through the tag array and the direction tag. Preferably, when a person passes through the LOS path, according to the diffraction model, the phase value of the tag will first increase and then decrease, and the curve in this interval is represented by C LOS , then calculate C LOS The area enclosed by the coordinate axis phase = 0:

[0040]

[0041] Among them, t start and t end It is C LOSThe intersection of phase = 0 corresponds to the time t of half the area. dir It can be expressed as follows:

[0042]

[0043] By comparing the direction label with the label immediately adjacent to it in the label array, dir To determine the direction of a person's walking: If the direction label t dir If it is smaller, it means that the walking direction is from the direction label to the label array, and vice versa.

[0044] Step S13. Obtain sample data of the sample object, and evaluate the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain the distinguishing ability value and anti-interference ability value of each tag. The spatial diversity of the tag array makes some tags have strong distinguishing ability for some people, while some tags have strong anti-interference ability for gait factors. These two types of tags should be assigned higher weights when participating in identity recognition. However, for different people and different gait factors, the distinguishing ability and anti-interference ability of the tag will also be different. Therefore, through the distinguishing ability evaluator and the anti-interference ability evaluator, the distinguishing ability and anti-interference ability of the tag are quantified in the training and recognition stages respectively, and then multiple tags are fused together according to the quantified values ​​of these two abilities to achieve identity recognition.

[0045] Considering the possible speed changes of the same object passing through the reader and tag array at different times, the dynamic time warping algorithm (DTW) is used to calculate the similarity of the phase waveforms of the object to be identified and the sample object. Therefore, the DTW distance is used to evaluate the discrimination ability and anti-interference ability of each tag.

[0046] In a preferred implementation of this embodiment, the evaluation method of the distinguishing ability value includes: calculating two DTW distances for each label based on the sample data of the sample object: internal distance and external distance. The former refers to the average DTW distance of any two samples of the same person, and the latter refers to the average DTW distance of two samples of different people. Generally speaking, the smaller the internal distance of a label and the larger the external distance, the stronger its distinguishing ability. Therefore, the distinguishing ability of each label can be quantified according to the following formula to obtain the corresponding distinguishing ability value:

[0047]

[0048] in, represents the mth training sample of the i-th label of object k.

[0049] Through the above formula, we can get the local distinguishing ability value of each label for any two people; we can also further convert α i Normalize to obtain the global distinguishing ability value of each label for all sample objects.

[0050] In a preferred implementation of this embodiment, the anti-interference capability value evaluation method includes: obtaining the phase data P of the i-th tag of the object to be identified i Then, calculate P i And the DTW distance of each sample in the training set, get the average DTW distance for each person (k represents object k); find the minimum value among them If the gait factor of the object to be identified is different from that in the training set and the label i is less affected, then the phase waveform should be similar to that of the object to be identified in the training set, so When label i is greatly affected, it will be significantly different from its own training waveform, even if it is close to others. will also be a larger value. Therefore, The size of the tag is used to quantify the anti-interference ability of the tag, and the anti-interference ability value of tag i is β i The calculation formula is as follows:

[0051]

[0052] in, represents the mth training sample of the i-th label of object k, Indicates phase data P i The average of the DTW distances of each sample in the training set.

[0053] Step S14. Based on the distinguishing ability value and the anti-interference ability value, a multi-dimensional weighted dynamic time warping algorithm (Weighted Multi-dimensional Dynamic Time Warping, WMD-DTW) is used to evaluate the similarity between the walking-related data of the object to be identified and the sample data to obtain the identity information of the object to be identified.

[0054] In a preferred implementation of this embodiment, the multidimensional weighted dynamic time warping algorithm includes: based on the sample data and the walking-related data, obtaining a sequence alignment distance matrix that meets a preset window restriction condition; the window restriction condition is: the difference between the row number and the column number of each element in the matrix does not exceed a preset value; based on the sequence alignment matrix, combined with the distinguishing ability value and anti-interference ability value of each label, obtaining the multidimensional weighted dynamic time warping distance between the object to be identified and the sample; selecting the smallest multiple distances from the multidimensional weighted dynamic time warping distances and performing weighted calculation to obtain the similarity.

[0055] In a preferred implementation of this embodiment, the method also includes: calculating the similarity based on the global distinguishing ability value, and extracting multiple groups of high-similar sample data with the highest similarity to the walking-related data; calculating the similarity between the walking-related data and multiple groups of the high-similar sample data based on the local distinguishing ability value to obtain the most similar sample data; the sample object identity information corresponding to the most similar sample data is the identity information of the object to be identified.

[0056] Specifically, this embodiment makes two modifications to the traditional DTW algorithm to reduce the amount of computation and prevent overstretching during sequence alignment. First, when generating the sequence alignment distance matrix, for the element c in the matrix i,j (c i,j is an element p in both sequences i and q j The present invention adds a window limit, and only when the difference between i and j is not too large (≤30) will c be calculated. i,j This is because the fine-grained extraction of walking-related data makes the best paths in the DTW distance matrix basically located on the diagonal of the matrix, and the window restriction can effectively reduce the amount of DTW calculation. In addition, the traditional DTW algorithm has a lot of problems in calculating c. i,j When considering only p i and q j The difference between i and j is ignored, which often makes the DTW distance calculated by two phase waveforms with similar amplitudes but different shapes very small. Therefore, the difference between i and j is also taken into account and the DTW calculation formula is revised as follows:

[0057] DTW(P,Q)=∑ i,j (|p i -q j |+γ×|ij|);

[0058] Wherein, γ is a constant factor. Preferably, this embodiment selects a smaller γ value, because if the γ value is too large, it is not conducive to the telescopic alignment of DTW, and |p i -qj | is generally much smaller than |ij|. Taking these factors into consideration, the present invention sets γ to 0.02.

[0059] Furthermore, since the phase data of a person walking contains phase waveforms of multiple labels, it is easy to think of using the multi-dimensional DTW (MD-DTW) algorithm to compare the similarity of two sets of phase data:

[0060]

[0061] Among them, P i and Q i Respectively represent the phase waveform of the i-th label in the two sets of data, and N is the number of labels. However, this simple multi-label fusion algorithm sets the participation of each label in identity recognition to be the same, which seriously limits the full utilization of the spatial diversity of the label array.

[0062] Therefore, after evaluating the discrimination ability and anti-interference ability of each label, the multidimensional weighted DTW distance (Weighted MD-DTW, WMD-DTW) can be obtained:

[0063]

[0064] Among them, α i and β i They represent the discrimination ability and anti-interference ability of label i respectively.

[0065] For the phase data P to be identified, the present invention calculates the similarity between it and the sample of object k in the training set according to the following method:

[0066] First, the system calculates the WMD-DTW distances between P and all samples of object k, and then selects the smallest L distances and arranges them from small to large as follows: d k,1 <d k,1 <… <d k,l , the final similarity is calculated by a weighted average algorithm:

[0067]

[0068] Then, the similarity between the data to be identified and each object in the training set is calculated. For the two objects with the highest similarity, the system will recalculate their similarity using the local distinguishing ability value of the label instead of the global value, and the object with higher similarity (smaller d(P,k)) is considered to be the object to be identified.

[0069] like Figure 2As shown, the embodiment of the present invention also provides a flow chart of a multi-label spatial fusion identity recognition method, which can be simply described as follows: the original phase signal collected from the reader will first be processed by the signal preprocessing module for phase unwrapping, smoothing, etc., and then input into the walking detection module for motion detection and walking recognition to extract phase data related to human walking; the extracted phase data will be further processed in the feature extraction module, including more accurate phase waveform extraction and wavelet denoising; then the walking direction will be identified according to the direction label; finally, the processed phase data will enter the identity recognition module, and the distinguishing ability and anti-interference ability of the tag will be quantified by the distinguishing ability evaluator and the anti-interference ability evaluator, and then the similarity between the extracted data and the sample data in the training set will be calculated according to the designed multi-dimensional weighted dynamic time warping algorithm (WMD-DTW), so as to determine its identity.

[0070] In some embodiments, the method can be applied to a controller, and the electronic control unit is, for example, an ARM (Advanced RISC Machines) controller, an FPGA (Field Programmable Gate Array) controller, a SoC (System on Chip) controller, a DSP (Digital Signal Processing) controller, or an MCU (Microcontroller Unit) controller, etc. In some embodiments, the method can also be applied to a computer including components such as a memory, a storage controller, one or more processing units (CPUs), a peripheral interface, an RF circuit, an audio circuit, a speaker, a microphone, an input / output (I / O) subsystem, a display screen, other output or control devices, and an external port; the computer includes but is not limited to personal computers such as desktop computers, laptop computers, tablet computers, smart phones, smart TVs, and personal digital assistants (PDAs). In other embodiments, the method can also be applied to a server, which can be arranged on one or more physical servers according to various factors such as function and load, or can be composed of distributed or centralized server clusters.

[0071] It is worth mentioning that, compared with the existing WiFi identification system, the present invention only requires people to use the tag array and the reader antenna to achieve identity recognition that is resistant to gait factor interference, while the WiFi identification system requires people to walk at least 5 meters in a straight line to obtain sufficient gait characteristics, and is also easily affected by gait factors; the present invention can be more universally deployed in practical applications, for example, the RFID reader and antenna can be embedded in the LED lamp; the present invention is particularly suitable for home or office environments, for example, the system can be deployed at the door of each room in the home to complete the identification of objects entering and leaving the room, and realize personalized services and behavior analysis based on each room; for some scenarios where RFID has been deployed, such as factories and companies, the existing equipment can be directly reused to realize non-invasive identification and tracking of employees.

[0072] In summary, the present invention designs an adaptive multi-label spatial fusion algorithm: first, the distinguishing ability and anti-interference ability of the labels are quantitatively evaluated, and then the accurate identification of the identity of the person when the gait factors are inconsistent is realized through the Weighted Multi-dimensional Dynamic Time Warping (WMD-DTW) algorithm; the interference of the gait factor is eliminated by suppressing the influence of the labels whose signal characteristics have changed significantly, thereby maximizing the advantages of spatial diversity.

[0073] Embodiment 2

[0074] like Figure 3 As shown, an embodiment of the present invention proposes a structural schematic diagram of an identity recognition system with multi-tag spatial fusion, which includes: an original data acquisition module 31, which acquires original phase data through an RFID tag array and a reader / writer deployed in a traffic area; a walking data extraction module 32, which is used to preprocess the original phase data and extract walking-related data of the object to be identified; a tag evaluation module 33, which is used to acquire sample data of a sample object, and evaluate the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain the distinguishing ability value and anti-interference ability value of each tag; an identity recognition module 34, which uses a multi-dimensional weighted dynamic time warping algorithm to evaluate the similarity between the walking-related data of the object to be identified and the sample data based on the distinguishing ability value and the anti-interference ability value, so as to obtain the identity information of the object to be identified.

[0075] It should be noted that the modules provided in this embodiment are similar to the methods and implementation methods provided above, so they will not be described in detail. It should also be noted that it should be understood that the division of the various modules of the above system is only a division of logical functions. In actual implementation, they can be fully or partially integrated into a physical entity, or physically separated. And these modules can all be implemented in the form of software called by processing elements; they can also be all implemented in the form of hardware; some modules can also be implemented in the form of software called by processing elements, and some modules can be implemented in the form of hardware. For example, the identity recognition module can be a separately established processing element, or it can be integrated in a chip of the above system for implementation. In addition, it can also be stored in the memory of the above system in the form of program code, and called and executed by a processing element of the above system. The function of the above identity recognition module. The implementation of other modules is similar. In addition, these modules can be fully or partially integrated together, or they can be implemented independently. The processing element described here can be an integrated circuit with signal processing capabilities. In the implementation process, each step of the above method or each of the above modules can be completed by an integrated logic circuit of hardware in the processor element or instructions in the form of software.

[0076] For example, the above modules may be one or more integrated circuits configured to implement the above methods, such as one or more application specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field programmable gate arrays (FPGAs). For another example, when a module is implemented in the form of a processing element scheduling program code, the processing element may be a general-purpose processor, such as a central processing unit (CPU) or other processors that can call program code. For another example, these modules may be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0077] Embodiment 3

[0078] An embodiment of the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the identity recognition method of multi-label spatial fusion is implemented.

[0079] Those skilled in the art can understand that all or part of the steps of implementing the above-mentioned method embodiments can be completed by hardware related to the computer program. The aforementioned computer program can be stored in a computer-readable storage medium. When the program is executed, the steps of the above-mentioned method embodiments are executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk and other media that can store program codes.

[0080] Embodiment 4

[0081] like Figure 4 As shown, an embodiment of the present invention provides a structural schematic diagram of an electronic terminal. The electronic terminal provided in this embodiment includes: a processor 41, a memory 42, and a communicator 43; the memory 42 is connected to the processor 41 and the communicator 43 through a system bus and completes mutual communication, the memory 42 is used to store computer programs, the communicator 43 is used to communicate with other devices, and the processor 41 is used to run computer programs, so that the electronic terminal executes each step of the identity recognition method of multi-label space fusion as above.

[0082] The system bus mentioned above can be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus, etc. The system bus can be divided into an address bus, a data bus, a control bus, etc. For ease of representation, only one thick line is used in the figure, but it does not mean that there is only one bus or one type of bus. The communication interface is used to realize the communication between the database access device and other devices (such as clients, read-write libraries, and read-only libraries). The memory may include random access memory (RAM), and may also include non-volatile memory, such as at least one disk storage.

[0083] The above-mentioned processor can be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; it can also be a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field programmable gate array (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components.

[0084] In summary, the present invention provides a multi-tag spatial fusion identity recognition method, system, medium and terminal, deploys RFID reader antennas and tag arrays on both sides of the traffic area, and quantitatively evaluates the discrimination ability and anti-interference ability of the tags, and then uses a multi-dimensional weighted dynamic time warping algorithm (Weighted Multi-dimensional Dynamic Time Warping, WMD-DTW) to achieve accurate identification of people when gait factors are inconsistent; eliminates the interference of gait factors by suppressing the influence of tags with obvious changes in signal characteristics, maximizes the advantages of spatial diversity, has extremely high identity recognition accuracy, and the system has good robustness. Therefore, the present invention effectively overcomes the various shortcomings of the prior art and has a high industrial utilization value.

[0085] The above embodiments are merely illustrative of the principles and effects of the present invention, and are not intended to limit the present invention. Anyone familiar with the art may modify or alter the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or alterations made by a person of ordinary skill in the art without departing from the spirit and technical concept disclosed by the present invention shall still be covered by the claims of the present invention.

Claims

1. A multi-label spatial fusion identity recognition method, characterized in that: include: The original phase data is obtained through the RFID tag array and reader / writer deployed in the traffic area; Preprocessing the original phase data and extracting walking-related data of the object to be identified; Acquire sample data of a sample object, and evaluate the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain a distinguishing ability value and an anti-interference ability value of each tag; Based on the discrimination ability value and the anti-interference ability value, a multi-dimensional weighted dynamic time warping algorithm is used to evaluate the similarity between the walking-related data of the object to be identified and the sample data, so as to obtain the identity information of the object to be identified; The evaluation method of the distinguishing ability value includes: Based on the sample data of the sample object, calculating the dynamic time warping distance of each tag, including the internal distance and the external distance; Normalizing the internal distance and the external distance to obtain a global distinguishing ability value of each label for all sample objects; obtaining a local distinguishing ability value of each label for every two sample objects based on the internal distance and the external distance; The evaluation method of the anti-interference capability value includes: After obtaining the phase data P of the i-th tag of the object to be identified i Then, calculate P i And the DTW distance of each sample in the training set, get the average DTW distance for each person (k represents object k); find the minimum value among them pass The size of the tag is used to quantify the anti-interference ability of the tag, and the anti-interference ability value of tag i is β i The calculation formula is as follows: in, represents the mth training sample of the i-th label of object k, Indicates phase data P i The average DTW distance of each sample in the training set; The multidimensional weighted dynamic time warping algorithm includes: Based on the sample data and the walking-related data, a sequence alignment distance matrix satisfying a preset window restriction condition is obtained; the window restriction condition is that the difference between the row number and the column number of each element in the matrix does not exceed a preset value; Based on the sequence alignment distance matrix, combined with the discrimination ability value and anti-interference ability value of each tag, a multi-dimensional weighted dynamic time warping distance between the object to be identified and the sample is obtained; The minimum multiple distances are selected from the multi-dimensional weighted dynamic time warping distances and weightedly calculated to obtain the similarity.

2. The identity recognition method according to claim 1, characterized in that: The distinguishing capability value includes: a global distinguishing capability value and a local distinguishing capability value; the method further includes: Calculating the similarity based on the global distinguishing ability value, and extracting multiple groups of highly similar sample data with the highest similarity to the walking-related data; Based on the local distinguishing ability value, the similarity between the walking-related data and the multiple groups of highly similar sample data is calculated respectively to obtain the most similar sample data; the sample object identity information corresponding to the most similar sample data is the identity information of the object to be identified.

3. The identity recognition method according to claim 1, characterized in that: The walking-related data includes walking direction data; The acquisition method includes: collecting the directional phase signal of the directional tag deployed on one side of the tag array; based on the directional phase signal, acquiring the time when the object to be identified passes through the tag array and the line-of-sight propagation path of the directional tag to obtain the walking direction data of the object to be identified.

4. The identity recognition method according to claim 1, characterized in that: include: Based on the line-of-sight propagation path of the tag array and the reader / writer, feature extraction is performed on the pre-processed raw phase data to obtain the walking-related data.

5. The identity recognition method according to claim 1, characterized in that: The method for acquiring the walking related data includes: Detecting whether an action occurs near the tag array based on the relative entropy of the original phase data, and extracting corresponding action data; Based on the time-frequency characteristics of the motion data, it is determined whether the motion that occurs is a walking motion through the traffic area, and corresponding walking-related data is extracted.

6. A multi-label spatial fusion identity recognition system, characterized in that: include: The raw data acquisition module acquires raw phase data through the RFID tag array and reader / writer deployed in the traffic area; A walking data extraction module, used to pre-process the original phase data and extract walking related data of the object to be identified; A tag evaluation module, used to obtain sample data of a sample object, and evaluate the distinguishing ability and anti-interference ability of each tag in the tag array in combination with the walking-related data of the object to be identified, so as to obtain a distinguishing ability value and an anti-interference ability value of each tag; An identity recognition module, based on the distinguishing ability value and the anti-interference ability value, uses a multi-dimensional weighted dynamic time warping algorithm to evaluate the similarity between the walking-related data of the object to be recognized and the sample data, so as to obtain the identity information of the object to be recognized; The evaluation method of the distinguishing ability value includes: Based on the sample data of the sample object, calculating the dynamic time warping distance of each tag, including the internal distance and the external distance; Normalizing the internal distance and the external distance to obtain a global distinguishing ability value of each label for all sample objects; obtaining a local distinguishing ability value of each label for every two sample objects based on the internal distance and the external distance; The evaluation method of the anti-interference capability value includes: After obtaining the phase data P of the i-th tag of the object to be identified i Then, calculate P i And the DTW distance of each sample in the training set, get the average DTW distance for each person (k represents object k); find the minimum value among them pass The size of the tag is used to quantify the anti-interference ability of the tag, and the anti-interference ability value of tag i is β i The calculation formula is as follows: in, represents the mth training sample of the i-th label of object k, Indicates phase data P i The average DTW distance of each sample in the training set; The multidimensional weighted dynamic time warping algorithm includes: Based on the sample data and the walking-related data, a sequence alignment distance matrix satisfying a preset window restriction condition is obtained; the window restriction condition is that the difference between the row number and the column number of each element in the matrix does not exceed a preset value; Based on the sequence alignment distance matrix, combined with the discrimination ability value and anti-interference ability value of each tag, a multi-dimensional weighted dynamic time warping distance between the object to be identified and the sample is obtained; The minimum multiple distances are selected from the multi-dimensional weighted dynamic time warping distances and weightedly calculated to obtain the similarity.

7. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the identity recognition method of multi-label spatial fusion described in any one of claims 1 to 5 is implemented.

8. An electronic terminal, characterized in that: include: Processor and memory; The memory is used to store a computer program, and the processor is used to execute the computer program stored in the memory, so that the terminal executes the identity recognition method of multi-label spatial fusion as described in any one of claims 1 to 5.

Images