A method, device and related equipment for gait recognition
By acquiring and quantifying the Hu moment feature data of the target object, a preset data model library for gait features is constructed, which solves the difficulty of identity recognition when the camera cannot capture a clear face, and realizes an efficient gait recognition method and improves identity Accuracy of identification.
Patent Information
- Application Number
- CN202110961394.7
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2021-08-20
- Publication Date
- 2024-12-03
- Estimated Expiration
- 2041-08-20
AI Technical Summary
When the camera cannot capture a clear face, how to improve the accuracy of identity recognition and reduce the problem of identification obstacles and difficulty in obtaining evidence.
By obtaining the Hu moment feature data in the image data of the target object, quantizing the code book, generating a sequence of numerical symbols that can be recognized by the computer as gait feature data, and matching it with the standard identification data in the preset data model library , use recurrent neural network for training to build a preset data model library of gait features to identify the gait of the target object.
When a clear face cannot be obtained, the accuracy of identity recognition is improved through gait recognition methods, the obstacles to identification and difficulties in obtaining evidence can be reduced, and the accurate identification of human identity can be achieved.
Smart Images

Figure CN113869109B_ABST
Abstract
Description
Technical Field
[0001] The present invention belongs to the technical field of gait recognition, and in particular relates to a gait recognition method, device and related equipment. Background Art
[0002] At present, most of the world's public security maintenance relies on video surveillance, and most of them use cameras to recognize people's faces to identify objects for judgment; gait recognition technology is an emerging biometric recognition technology that aims to identify people through their walking posture. Gait recognition has the advantages of long distance, cross-viewing angle, uncontrolled, non-contact and not easy to disguise. In the field of intelligent video surveillance, gait recognition technology is particularly important, especially when the camera cannot capture a clear face; for example, in a dim environment, the camera cannot accurately capture the face of the person. If the person needs to be identified, it will cause great recognition obstacles and difficulty in obtaining evidence;
[0003] It can be seen that when the camera cannot capture a clear face and the person needs to be identified, it will cause great obstacles to identification and difficulty in obtaining evidence;
[0004] That is, when the camera cannot capture a clear face, how to improve the accuracy of human identity recognition, thereby reducing obstacles to human identification and difficulty in obtaining evidence, is a technical problem that urgently needs to be solved in this field. Summary of the invention
[0005] The technical problem to be solved by the present invention is how to provide a method, device and related equipment for gait recognition to at least solve some of the above technical problems.
[0006] In order to solve at least some of the above technical problems, in a first aspect, the present invention provides a method for gait recognition, which is used to identify whether a target object is an identification object by walking posture, and the method comprises: obtaining basic identification data of the target object; the obtaining basic identification data of the target object comprises: obtaining image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book;
[0007] The basic identification data is identified and matched with the standard identification data in the preset data model library; the method for constructing the preset data model library includes: obtaining the identification data of the identified object, the obtaining of the identification data of the identified object includes obtaining the image data of the identified object; extracting the second Hu moment feature data in the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment feature data of multiple periods of the step posture, thereby generating a codebook, and the second Hu moment feature data is quantized into a set of digital symbol sequences that can be recognized by a computer according to the codebook, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the identified object is quantized into digital symbols through the codebook, and the recurrent neural network is trained; the trained recurrent neural network is obtained to characterize the gait characteristics of the identified object according to the preset data model library.
[0008] If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
[0009] In the first aspect, before extracting the first Hu moment feature data from the image data of the target object, the method further includes: performing segmentation processing on the image of the target object to separate the target object from the background in the image.
[0010] In the first aspect, segmenting the image of the target object includes: separating the image of the target object from the background using a median filter.
[0011] In a second aspect, the present invention provides a device for gait recognition, which is used to identify an identified target object through the walking posture of the target object, and the device includes: a basic data acquisition module: acquiring basic identification data of the target object; the acquisition of the basic identification data of the target object includes: acquiring image data of the target object, extracting first Hu moment feature data in the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book; a matching module, matching the basic identification data with standard identification data in a preset data model library; the method for constructing the preset data model library includes: acquiring identification data of the identified object, and the acquisition of identification data of the identified object includes, Acquire the image data of the identified object; extract the second Hu moment feature data from the image data of the identified object; calculate the period of the step posture of the identified object based on the second Hu moment feature data; calculate the average value of each second Hu moment value of the multiple steps and postures, so as to generate a codebook, and quantize the second Hu moment feature data into a set of digital symbol sequences that can be recognized by the computer according to the codebook, and the digital symbol sequence is used as the input data for recurrent neural network training; the step posture feature data of the identified object is quantized into digital symbols through the codebook, and the recurrent neural network is trained; the trained recurrent neural network is used to characterize the preset data model library of the gait characteristics of the identified object. The judgment module judges that the target object is the identified object if the basic identification data is recognized by the recurrent neural network in the preset data model library.
[0012] In the second aspect, the basic data acquisition module is further used to: perform segmentation processing on the image of the target object to separate the target object from the background in the image.
[0013] In the second aspect, the basic data acquisition module is further used to separate the image of the target object from the background using median filtering.
[0014] In a third aspect, the present invention provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is used to implement the following steps when executing the computer program stored in the memory: obtaining basic identification data of a target object; obtaining the basic identification data of the target object comprises: obtaining image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences recognizable by a computer as the gait feature data according to a code book; matching the basic identification data with standard identification data in a preset data model library; and constructing a preset data model library comprises: obtaining identification data of an identified object, obtaining the identification data of an identified object, The recognition data of an object includes: obtaining image data of the object to be identified; extracting second Hu moment feature data from the image data of the object to be identified; calculating the period of the step posture of the object to be identified based on the second Hu moment feature data; calculating the average value of each second Hu moment value of a plurality of the periods of the step postures, thereby generating a codebook, wherein the second Hu moment feature data is quantized into a set of digital symbol sequences that can be recognized by a computer according to the codebook, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the object to be identified is quantized into digital symbols through the codebook, and the recurrent neural network is trained; the preset data model library for the trained recurrent neural network to characterize the gait characteristics of the identified object is obtained;
[0015] If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
[0016] In the third aspect, the processor may further implement the following steps when used to execute the computer program stored in the memory: segmenting the image of the target object so as to separate the target object from the background in the image.
[0017] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program implements the following steps when executed by a processor: obtaining basic identification data of a target object; the obtaining basic identification data of a target object comprises: obtaining image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences recognizable by a computer as the gait feature data according to a code book; matching the basic identification data with standard identification data in a preset data model library; a method for constructing the preset data model library comprises: obtaining identification data of an identified object, the obtaining identification data of an identified object comprises: obtaining The invention relates to a method for obtaining a step posture period of the identified object by extracting the second Hu moment feature data from the image data of the identified object; calculating the period of the step posture of the identified object according to the second Hu moment feature data; calculating the average value of each second Hu moment value of the period of the step posture of a plurality of the steps, thereby generating a codebook, wherein the second Hu moment feature data is quantized into a set of digital symbol sequences that can be recognized by a computer according to the codebook, and the digital symbol sequence is used as input data for training a recurrent neural network; the step posture feature data of the identified object is quantized into digital symbols through the codebook, and the recurrent neural network is trained; and the preset data model library for characterizing the gait characteristics of the identified object is obtained by using the trained recurrent neural network;
[0018] If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
[0019] Beneficial effects:
[0020] A gait recognition method provided by the present invention identifies an identification object through the walking posture of the target object. During the identification, basic identification data of the target object is first obtained, and then the basic identification data is matched with standard identification data in a preset data model library. If the match is successful, the target object is judged to be an identification object. If the match is unsuccessful, the target object is judged to be a non-identification object. The above method is used to achieve the situation when the camera cannot capture a clear face and the person needs to be identified. In the identification stage, in order to obtain the image features of the target object more accurately, the first Hu moment feature data in the image data is collected, and in the learning stage, in order to To make the acquisition of the image data of the target object more accurate, the second Hu moment feature data in the image data is collected, and then the second Hu moment feature data is used to calculate the period of the step posture of the identified object; the average value of each of the second Hu moment values of the multiple steps and postures is calculated to generate a codebook, and the Hu moment feature data is quantified into a set of digital symbol sequences according to the codebook as the input of the next recurrent neural network training, and one gait has one codebook; the step posture feature data of the identified object is quantified into digital symbols through the codebook, and the recurrent neural network is trained to obtain a database for calling when being identified. The technical effect of identifying the target object by the above gait recognition method.
[0021] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented according to the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0022] In order to more clearly illustrate the embodiments of this specification or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0023] Figure 1 Schematic diagram of different states of the target object during the segmentation and extraction process of the present invention;
[0024] Figure 2 is a flowchart of a method for gait recognition in Embodiment 1 of the present invention;
[0025] Figure 3 This is a flow chart of recurrent neural network training in the present invention;
[0026] Figure 4This is a flow chart of recurrent neural network recognition in an embodiment of the present invention;
[0027] Figure 5 This is a flow chart of an implementation of a recurrent neural network in Example 5 of the present invention;
[0028] Figure 6 This is a schematic diagram of the structure of an electronic device in Embodiment 3 of the present invention;
[0029] Figure 7 It is a schematic diagram of the structure of four computer-readable storage media in accordance with an embodiment of the present invention;
[0030] Description of reference numerals:
[0031] 300. Electronic equipment;
[0032] 310. Memory;
[0033] 311. Computer programs;
[0034] 320. Processor;
[0035] 400. Computer readable storage medium;
[0036] 411. Computer programs. DETAILED DESCRIPTION
[0037] The following will be combined with the drawings in the embodiments of this specification to clearly and completely describe the technical solutions in the embodiments of this specification. Obviously, the described embodiments are only part of the embodiments of this specification, not all of the embodiments. Based on the embodiments in this specification, all other embodiments obtained by ordinary technicians in this field belong to the scope of protection of the present invention.
[0038] At the same time, in the embodiments of this specification, when a component is referred to as being "fixed to" another component, it may be directly on the other component or there may also be a centered component. When a component is considered to be "connected to" another component, it may be directly connected to the other component or there may be a centered component at the same time. When a component is considered to be "set on" another component, it may be directly set on the other component or there may be a centered component at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used in the embodiments of this specification are for illustrative purposes only and are not intended to limit the present invention.
[0039] Embodiment 1:
[0040] like Figure 1-7As shown, a first embodiment of the present invention provides a gait recognition method for identifying whether a target object is an identification object by walking posture, the method comprising: obtaining basic identification data of the target object; the obtaining basic identification data of the target object comprises: obtaining image data of the target object, extracting first Hu moment feature data in the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book; matching the basic identification data with standard identification data in a preset data model library; the method for constructing the preset data model library comprises: obtaining identification data of the identified object, the obtaining identification data of the identified object comprises: obtaining image data of the identified object; extracting the first Hu moment feature data of the identified object; The second Hu moment feature data in the image data; the period of the step posture of the identified object is calculated based on the second Hu moment feature data; the average value of each second Hu moment value of the multiple steps posture periods is calculated to generate a code book, and the Hu moment feature data is quantized into a set of digital symbol sequences according to the code book as the input of the next recurrent neural network training, and one gait has one code book; the recurrent neural network is trained after the step posture feature data of the identified object is quantized into digital symbols through the code book; the preset data model library of the trained recurrent neural network is obtained to characterize the gait characteristics of the identified object; if the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is judged to be the identified object.
[0041] In the above technical scheme, the present invention provides a method for gait recognition, which identifies the identification object through the walking posture of the target object. During the identification, the basic identification data of the target object is first obtained, and then the basic identification data is matched with the standard identification data in the preset data model library. If the match is successful, the target object is judged to be the identification object. If the match is unsuccessful, the target object is judged to be a non-identification object. Through the above method, when the camera cannot capture a clear face and it is necessary to identify the person, in the identification stage, in order to make the image features of the target object more accurate, the first Hu moment feature data in the image data is collected, and in the learning stage, In order to obtain the image data of the target object more accurately, the second Hu moment feature data in the image data is collected, and then the second Hu moment feature data is used to calculate the period of the step posture of the identified object; the average value of each of the second Hu moment values of the multiple steps and postures is calculated to generate a codebook, and the Hu moment feature data is quantified into a set of digital symbol sequences according to the codebook as the input of the next recurrent neural network training, and one gait has one codebook; the step posture feature data of the identified object is quantified into digital symbols through the codebook, and the recurrent neural network is trained; the trained recurrent neural network is obtained for the database called when it is identified. The technical effect of identifying the target object by the above gait recognition method.
[0042] Specifically, as for the steps of constructing the preset data model library, it can be understood that this step is a preparatory step for identifying the identified object. The identification data of the target object is first stored and trained, for example, the image information of the step posture of a person walking is feature extracted, and then the training is stored as a program that can be recognized by a computer to form a preset data model library. It includes the following steps: first, the identification data of the identified object is obtained, and the identification data of the identified object is obtained, and then the step posture feature data of the identified object when moving is extracted based on the image data; the step posture feature data is quantified into digital symbols through the code book, and the recurrent neural network is trained; the trained recurrent neural network is used to characterize the preset data model library of the gait characteristics of the identified object.
[0043] Based on the above embodiment, for a recurrent neural network, it can be interpreted as a network trained by a computer program recognition, and those skilled in the art can understand that: the current output of a sequence is also related to the previous output. The specific manifestation is that the network will remember the previous information and apply it to the calculation of the current output, that is, the nodes between the hidden layers are no longer disconnected but connected, and the input of the hidden layer includes not only the output of the input layer but also the output of the hidden layer at the previous moment.
[0044] As for the steps of obtaining the gait feature data of the target object when it moves based on the image data, it specifically includes the following implementation methods: extracting the first Hu moment feature data in the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book; Hu moment has the characteristics of translation invariance, rotation invariance, and scale invariance in shape matching, which can greatly improve the robustness of the segmented gait feature description and the performance of feature matching.
[0045] Based on the above-mentioned robustness, those skilled in the art can know that robustness is the transliteration of Robust, which means strong and strong. It is the key to the survival of the system in abnormal and dangerous situations. For example, whether a computer software can survive an input error, disk failure, network overload or intentional attack without freezing or crashing is the robustness of the software. The so-called "robustness" refers to the characteristic of a control system maintaining certain other performances under certain (structure, size) parameter perturbations. According to different definitions of performance, it can be divided into stability robustness and performance robustness. A fixed controller designed with the robustness of a closed-loop system as the goal is called a robust controller.
[0046] Based on the step of extracting the first Hu moment feature data from the image data of the target object in the above embodiment, the following method step is also included before this step: segmenting the image of the target object to separate the target object from the background in the image.
[0047] Furthermore, the segmentation process of the image of the target object specifically includes: using median filtering to separate the image of the target object from the background. It can be understood that median filtering has a good filtering effect on impulse noise, especially while filtering noise, it can protect the edge of the signal so that it is not blurred.
[0048] Further, please participate Figure 1 Figure (a) is the original image of the target object, Figure (b) is the image after segmentation, that is, the image in which the person is separated from the background; Figure (c) is the segmented image after median filtering; Figure (d) is the image after binarization. It can be clearly seen that from images (a) to (d), the segmentation effect of each image changes progressively, that is, the background impurities extracted from images (a) to (d) are gradually removed.
[0049] In the recognition stage, extracting the step posture features of the identified object when it moves based on the image data includes: extracting the second Hu moment feature data from the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each of the second Hu moment values of multiple steps and posture periods, and generating the step posture feature data that can be recognized by the computer based on the average value, which can also be understood as: calculating the average value of the Hu moments of the corresponding images within several gait cycles to generate a code book. The Hu moments of other training image data are quantized into a set of symbol sequences according to the code book as the input of the next recurrent neural network training, and one gait has one code book.
[0050] Embodiment 2:
[0051] Embodiment 2 of the present invention provides a gait recognition device, which is used to identify a target object to be identified through the walking posture of the target object. The device includes: a basic data acquisition module, a matching module and a judgment module. The basic data acquisition module is used to acquire basic identification data of the target object; the acquisition of the basic identification data of the target object includes: acquiring image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer according to a code book as the gait feature data; the matching module is used to match the basic identification data with standard identification data in a preset data model library; the method for constructing the preset data model library includes: acquiring the basic identification data of the identified object Identification data, the acquisition of identification data of the identified object includes, acquiring the image data of the identified object; extracting the second Hu moment feature data in the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment value of the multiple steps and posture periods, and generating the step posture feature data that can be recognized by the computer based on the average value; the step posture feature data of the identified object is used to train the recurrent neural network; the trained recurrent neural network is obtained to characterize the preset data model library of the gait characteristics of the identified object; if the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is judged to be the identified object. In this way, when the camera cannot capture a clear face and it is necessary to identify a person, the technical effect of identifying a person is achieved through the device of the second embodiment of the present invention.
[0052] Specifically, for the basic data acquisition module, as an implementation method of the basic data module, it may include the following steps: first extract the first Hu moment feature data in the image data of the target object; then obtain the gait feature data based on the first Hu moment feature data. Those skilled in the art can know that the Hu moment has the characteristics of translation invariance, rotation invariance, and scale invariance in shape matching, which can greatly improve the robustness of the segmented gait feature description and the performance of feature matching.
[0053] Please refer to Figure 2 , step S1 is used to characterize: obtaining basic identification data of the target object; step S2 is used to characterize: matching the basic identification data with standard identification data in a preset data model library; if they match, step S3 is used to characterize: judging that the target object is an identification object.
[0054] Embodiment three:
[0055] like Figure 6 As shown, the third embodiment of the present invention provides an electronic device 300, including: a memory 310, a processor 320, and a computer program 311 stored in the memory 310 and executable on the processor 320, wherein the processor 320 is used to implement the following steps when executing the computer program 311 stored in the memory 310: obtaining basic identification data of the target object; the obtaining basic identification data of the target object includes: obtaining image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book; matching the basic identification data with standard identification data in a preset data model library; the method for constructing the preset data model library includes: obtaining identification data of the identified object; According to the method, the identification data of the identified object is obtained, including: obtaining image data of the identified object; extracting second Hu moment feature data from the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment value of multiple periods of the step posture, thereby generating a code book, and quantizing the Hu moment feature data into a set of digital symbol sequences as input for recurrent neural network training according to the code book; training the recurrent neural network after the step posture feature data of the identified object is quantized into digital symbols through the code book; obtaining the preset data model library for the trained recurrent neural network to characterize the gait characteristics of the identified object; if the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be the identified object.
[0056] Embodiment 4:
[0057] like Figure 7 As shown, embodiment 4 of the present invention provides a computer-readable storage medium 400, on which a computer program 411 is stored, and when the computer program 411 is executed by a processor, the following steps are implemented: obtaining basic identification data of the target object; the obtaining basic identification data of the target object includes: obtaining image data of the target object, extracting first Hu moment feature data in the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as the gait feature data according to a code book; matching the basic identification data with standard identification data in a preset data model library; the method for constructing the preset data model library includes: obtaining identification data of the identified object, the obtaining identification data of the identified object includes obtaining The invention relates to a method for obtaining image data of an object to be identified; extracting second Hu moment feature data from the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment value of multiple periods of the step postures, thereby generating a code book, and quantizing the Hu moment feature data into a set of digital symbol sequences according to the code book as input for recurrent neural network training; training the recurrent neural network after the step posture feature data of the identified object is quantized into digital symbols through the code book; obtaining the preset data model library for the trained recurrent neural network to characterize the gait characteristics of the identified object; if the basic identification data is recognized by the recurrent neural network in the preset data model library, then judging that the target object is the identified object.
[0058] Embodiment five:
[0059] See also Figure 5 The embodiment of the present invention adopts an N vs 1 recurrent neural network (RNNs) structure, that is, the input is a sequence of N symbols, but the output is a single value. This structure is particularly suitable for sequence classification problems. Here, we input the gait Hu moment quantized into a sequence of digital symbols according to the code book, and the output is a number that specifies the individual's identity number. Figure 5 The diagram shows a 5vs1 recurrent neural network (RNN) where the input is a sequence of 5 symbols and the output is a single value.
[0060] In several embodiments provided by the present invention, it should be understood that the disclosed devices and methods can also be implemented in other ways. The device embodiments described above are merely schematic. For example, the flowcharts and block diagrams in the accompanying drawings show the possible architecture, functions and operations of the devices, methods and computer program products according to multiple embodiments of the present invention. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the above-mentioned module, a program segment or a part of a code contains one or more executable instructions for implementing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two consecutive boxes can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or action, or can be implemented with a combination of dedicated hardware and computer instructions.
[0061] For another example, the division of the above units is only a logical function division, and there may be other division methods in actual implementation. For another example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some communication interface, device or unit, which can be electrical, mechanical or other forms.
[0062] The units described above as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0063] In addition, each functional unit in the embodiment provided by the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0064] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the above methods of each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc., various media that can store program codes.
[0065] It should be noted that similar numbers and letters represent similar items in the following figures. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", etc. are only used to distinguish the description and cannot be understood as indicating or implying relative importance.
[0066] Finally, it should be noted that the above embodiments are only specific implementations of the present invention, which are used to illustrate the technical solutions of the present invention, rather than to limit them. The protection scope of the present invention is not limited thereto. Although the present invention is described in detail with reference to the above embodiments, ordinary technicians in the field should understand that any technician familiar with the technical field can still modify the technical solutions recorded in the above embodiments within the technical scope disclosed by the present invention, or can easily think of changes, or make equivalent replacements for some of the technical features therein; and these modifications, changes or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention. They should all be covered within the protection scope of the present invention. Therefore, the protection scope of the present invention shall be based on the protection scope of the claims.
Claims
1. A gait recognition method for identifying whether a target object is an identification object by walking posture, characterized in that: The method comprises: Acquire basic identification data of the target object; the acquiring basic identification data of the target object comprises: acquiring image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as gait feature data according to a code book; The basic identification data is identified and matched with the standard identification data in the preset data model library; the method for constructing the preset data model library includes: obtaining the identification data of the identified object, the obtaining of the identification data of the identified object includes obtaining the image data of the identified object; extracting the second Hu moment feature data in the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment feature data of multiple periods of the step posture, thereby generating a code book, and the second Hu moment feature data is quantified into a set of digital symbol sequences that can be recognized by the computer according to the code book, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the identified object is quantized into digital symbols through the code book, and the recurrent neural network is trained; the preset data model library obtained by the trained recurrent neural network is used to characterize the gait characteristics of the identified object; If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
2. The method for gait recognition according to claim 1, characterized in that: Before extracting the first Hu moment feature data from the image data of the target object, the method further includes: The image of the target object is segmented to separate the target object from the background in the image.
3. The method for gait recognition according to claim 1, characterized in that: The segmentation process of the image of the target object includes: Use median filtering to separate the image of the target object from the background.
4. A gait recognition device, used to identify a target object by its walking posture, characterized in that: The device comprises: Basic data acquisition module: acquire basic identification data of the target object; the acquisition of the basic identification data of the target object includes: acquiring image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as input data for recurrent neural network training according to a code book; A matching module identifies and matches the basic identification data with the standard identification data in a preset data model library; a method for constructing the preset data model library includes: obtaining identification data of an identified object, wherein obtaining identification data of an identified object includes obtaining image data of the identified object; extracting second Hu moment feature data from the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment feature data of a plurality of the steps and postures, thereby generating a code book, wherein the second Hu moment feature data is quantified into a set of digital symbol sequences that can be recognized by a computer according to the code book, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the identified object is quantified into digital symbols through the code book, and the recurrent neural network is trained; the preset data model library in which the trained recurrent neural network is used to characterize the gait characteristics of the identified object is obtained; The judgment module judges that the target object is an identification object if the basic identification data is recognized by a recurrent neural network in a preset data model library.
5. The gait recognition device according to claim 4, characterized in that: The basic data acquisition module is also used for: The image of the target object is segmented to separate the target object from the background in the image.
6. The gait recognition device according to claim 4, characterized in that: The basic data acquisition module is also used for: Use median filtering to separate the image of the target object from the background.
7. An electronic device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor is configured to implement the following steps when executing the computer program stored in the memory: Acquire basic identification data of the target object; the acquiring basic identification data of the target object comprises: acquiring image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as gait feature data according to a code book; Match the basic identification data with the standard identification data in the preset data model library; the method for constructing the preset data model library includes: obtaining the identification data of the identified object, the obtaining the identification data of the identified object includes obtaining the image data of the identified object; extracting the second Hu moment feature data in the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment feature data of multiple steps and postures, thereby generating a code book, and quantizing the second Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer according to the code book, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the identified object is quantized into digital symbols through the code book, and the recurrent neural network is trained; the preset data model library obtained by the trained recurrent neural network is used to characterize the gait characteristics of the identified object; If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
8. The electronic device according to claim 7, characterized in that: The processor may further implement the following steps when used to execute the computer program stored in the memory: The image of the target object is segmented to separate the target object from the background in the image.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the following steps are implemented: Acquire basic identification data of the target object; the acquiring basic identification data of the target object comprises: acquiring image data of the target object, extracting first Hu moment feature data from the image data of the target object; quantizing the first Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer as gait feature data according to a code book; Match the basic identification data with the standard identification data in the preset data model library; the method for constructing the preset data model library includes: obtaining the identification data of the identified object, the obtaining the identification data of the identified object includes obtaining the image data of the identified object; extracting the second Hu moment feature data in the image data of the identified object; calculating the period of the step posture of the identified object based on the second Hu moment feature data; calculating the average value of each second Hu moment feature data of multiple steps and postures, thereby generating a code book, and quantizing the second Hu moment feature data into a set of digital symbol sequences that can be recognized by a computer according to the code book, and the digital symbol sequence is used as input data for recurrent neural network training; the step posture feature data of the identified object is quantized into digital symbols through the code book, and the recurrent neural network is trained; the preset data model library obtained by the trained recurrent neural network is used to characterize the gait characteristics of the identified object; If the basic identification data is recognized by the recurrent neural network in the preset data model library, the target object is determined to be an identification object.
Citation Information
Patent Citations
Active detection and recognition system and method for human gait behavior based on semantic folding
CN107423730A
Gait information recognition optimization method and system and storage medium
CN111144167A
Target object gait processing method and device and nonvolatile storage medium
CN112597881A
Monitoring activity using video information
US20060018516A1
