CN114612934A - Gait sequence evaluation method and system based on quality dimension - Google Patents
Gait sequence evaluation method and system based on quality dimension Download PDFInfo
- Publication number
- CN114612934A CN114612934A CN202210249668.4A CN202210249668A CN114612934A CN 114612934 A CN114612934 A CN 114612934A CN 202210249668 A CN202210249668 A CN 202210249668A CN 114612934 A CN114612934 A CN 114612934A
- Authority
- CN
- China
- Prior art keywords
- gait
- sequence
- quality
- stage
- gait sequence
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Health & Medical Sciences (AREA)
- Software Systems (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Mathematical Physics (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses a gait sequence evaluation method and a system based on quality dimension, wherein the method comprises the steps of obtaining a single-stage gait sequence; acquiring a first characteristic parameter of a single-stage gait sequence; evaluating the quality of the single-stage gait sequence according to the first characteristic parameter to obtain the quality fraction of the single-stage gait sequence; acquiring a multi-section gait sequence; acquiring second characteristic parameters of the multi-section gait sequence; evaluating the quality of the multi-segment gait sequence according to the quality fraction of the single-segment gait sequence and the second characteristic parameter to obtain the quality fraction of the multi-segment gait sequence; and comparing the mass fraction of the single-stage gait sequence or the mass fraction of the multi-stage gait sequence with a corresponding preset fraction threshold, and storing the single-stage gait sequence or the multi-stage gait sequence in a gait base database in a grading manner according to a comparison result. The invention can quickly and accurately acquire the quality of the gait sample, is convenient for hierarchical storage, and can use the gait samples with different qualities aiming at different application scenes.
Description
Technical Field
The invention discloses a gait sequence evaluation method and system based on quality dimension, and belongs to the technical field of gait recognition.
Background
Gait is the biological and behavioral characteristics of natural human, and reflects the change rule of body shape in time and space. Because the gait information has the characteristics of high anti-counterfeiting performance, wearing span, visual angle span and the like, the gait identification method has higher accuracy by identifying the identity of a person through the gait.
In the process of identifying the identity of a person by using gait, the collected gait sample needs to be compared with the gait sample prestored in the gait database. Therefore, the quality of the gait sample pre-stored in the gait database will directly affect the accuracy of the identification result.
In the prior art, the gait recognition result accuracy is improved by increasing samples of people in a gait database under various external environments, but the data volume in the gait database is large due to the mode, and the condition of poor quality of part of data exists, so that the gait recognition result accuracy is low and the recognition speed is slow.
Disclosure of Invention
The application aims to provide a gait sequence evaluation method and system based on quality dimensionality, and the method and system are used for solving the technical problems of low accuracy and low recognition speed of gait recognition results caused by large quantity and poor quality of gait samples prestored in a gait database in the prior art.
The invention provides a gait sequence evaluation method based on quality dimension in a first aspect, which comprises the following steps:
acquiring a single-stage gait sequence;
acquiring a first characteristic parameter of the single-stage gait sequence;
and evaluating the quality of the single-stage gait sequence according to the first characteristic parameter to obtain the quality fraction of the single-stage gait sequence.
Preferably, the first characteristic parameter includes at least one of gait cycle quality, sharpness of human silhouette, illumination intensity and viewing angle.
Preferably, when the first characteristic parameter is the gait cycle quality, acquiring the first characteristic parameter of the single gait sequence specifically includes:
acquiring a gait silhouette sequence corresponding to the single-stage gait sequence;
determining the gait cycle quality of the gait silhouette sequence, which specifically comprises the following steps:
calculating the contact ratio between gait silhouette images in the gait silhouette sequence;
and counting the number of the contact ratios larger than a preset contact ratio threshold value, and determining the gait cycle quality of the gait silhouette sequence according to the number.
Preferably, the calculating the contact ratio between the gait silhouette images in the gait silhouette sequence specifically includes:
acquiring a plurality of gait silhouette images with set frame number intervals in a gait silhouette sequence;
respectively calculating the contact ratio between two adjacent gait silhouette images in the plurality of gait silhouette images, which specifically comprises the following steps:
sequentially acquiring two adjacent gait silhouette images from the multiple gait silhouette images;
and calculating the contact ratio between the two adjacent gait silhouette images according to the pixel values at the same pixel positions of the two adjacent gait silhouette images.
Preferably, when the first characteristic parameter is the definition of the silhouette, the acquiring the first characteristic parameter of the single gait sequence specifically includes:
converting each gait image in the single gait sequence into a gray image;
filtering the gray level image to obtain a filtering sequence;
and calculating the variance of each filtering image in the filtering sequence, and determining the definition of the human-shaped contour according to the variance.
Preferably, after acquiring the single-stage gait sequence, the method further comprises the following steps:
judging whether the single gait sequence meets a preset pre-evaluation condition or not; if yes, acquiring a first characteristic parameter of the single-stage gait sequence;
if not, deleting the single gait sequence;
wherein the pre-evaluation conditions comprise complete human shape, natural walking state and image quality;
the image quality comprises at least one of the resolution of the gait image, the size of the gait image and the frame number of the gait image;
correspondingly, whether the single-stage gait sequence meets a preset pre-evaluation condition is judged, and the method specifically comprises the following steps:
judging whether the gait images in the single-stage gait sequence contain complete humanoid figures or not;
judging whether the pedestrian in the single-stage gait sequence is in a natural walking state or not;
and judging whether the image quality of the gait image in the single-stage gait sequence is greater than a preset quality threshold value or not.
Preferably, the determining whether the pedestrian in the single-stage gait sequence is in a natural walking state specifically includes:
respectively acquiring the mean value, the standard deviation and the covariance of two adjacent gait images in the single-stage gait sequence;
calculating first similarities of two adjacent gait images according to the mean value, the standard deviation and the covariance to obtain a plurality of first similarities;
and calculating the average value of the plurality of first similarity, and judging whether the average value is greater than a preset similarity threshold value, if so, determining that the pedestrian in the single-stage gait sequence is in a static state.
Preferably, the method further comprises the following steps:
acquiring a multi-segment gait sequence, wherein the multi-segment gait sequence comprises a plurality of single-segment gait sequences;
acquiring second characteristic parameters of the multi-segment gait sequence;
evaluating the quality of the multiple gait sequences according to the quality fraction of the single gait sequence and the second characteristic parameter to obtain the quality fraction of the multiple gait sequences;
the second characteristic parameters include the number of sequences and the view richness.
Preferably, the method further comprises the following steps:
and comparing the mass fraction of the single gait sequence or the mass fractions of the multiple gait sequences with a corresponding preset fraction threshold, and storing the single gait sequence or the multiple gait sequences in a gait base database in a grading manner according to a comparison result.
A second aspect of the present invention provides a gait sequence evaluation system based on a quality dimension using the above gait sequence evaluation method based on a quality dimension, including:
the sequence acquisition module is used for acquiring a single-stage gait sequence;
the characteristic acquisition module is used for acquiring a first characteristic parameter of the single gait sequence;
and the evaluation module is used for evaluating the quality of the single gait sequence according to the first characteristic parameter to obtain the quality score of the single gait sequence.
Compared with the prior art, the gait sequence evaluation method and system based on the quality dimension have the following beneficial effects:
the method comprises a single-stage gait sequence evaluation method and a multi-stage gait sequence evaluation method combined with the single-stage gait sequence evaluation method. The evaluation method can quickly and accurately acquire the quality of the gait sequence, is convenient for hierarchical storage, and can use the gait sequences with different qualities aiming at different application scenes. For example, gait sequences with a quality greater than or equal to a preset quality threshold may be used for gait recognition; and using the gait sequence with the quality less than the preset quality threshold value for gait retrieval.
Drawings
FIG. 1 is a flow chart of a gait sequence assessment method based on quality dimensions according to an embodiment of the invention;
FIG. 2 is a sequence diagram of an unnormalized gait silhouette in an embodiment of the present invention;
FIG. 3 is a sequence diagram of a normalized gait silhouette in an embodiment of the invention;
FIG. 4 is a flow chart of a gait sequence assessment method based on quality dimensions according to another embodiment of the invention;
fig. 5 is a schematic structural diagram of a gait sequence evaluation system based on quality dimension according to an embodiment of the present invention.
In the figure, 101 is a sequence acquisition module; 102 is a characteristic obtaining module; 103 is an evaluation module.
Detailed Description
In the following description, for purposes of explanation and not limitation, specific details are set forth, such as particular system structures, techniques, etc. in order to provide a thorough understanding of the embodiments of the invention. It will be apparent, however, to one skilled in the art that the present invention may be practiced in other embodiments that depart from these specific details. In other instances, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.
Before explaining the technical solution of the present invention in detail, terms used in the present invention will be described first.
Gait: the biological characteristics and behavior characteristics of natural human body reflect the change rule of body figure in space and time.
Gait sample: the gait video or the gait image sequence (gait sequence for short) of the natural person is obtained by the modes of collection, pretreatment and the like. The sample in the invention refers to an original (cut) video or a continuous image sequence containing a natural human gait cycle, and contains information of clothes, shoes, hats and the like.
Gait silhouette sequence: the gait sample is a human body image represented by a solid shape of a single color (the background is black and the human shape is white generally) after background useless information is removed.
Gait cycle: when the walking stick is used for walking normally and continuously, the two feet respectively complete the advancing process from heel off-ground lifting to heel on-ground.
Gait recognition: the natural person is automatically identified based on the biological characteristics and behavioral characteristics of the natural person included in gait. The gait recognition can be used for identity recognition and can also be used in non-identity recognition scenes, such as behavior analysis, posture analysis or anomaly analysis.
A gait base bank: a database storing gait data may be ranked by mass fraction.
The technical solution of the present invention will be described in detail below.
The embodiment of the invention provides a gait sequence evaluation method based on quality dimension in a first aspect, which comprises a single-stage gait sequence evaluation method and a multi-stage gait sequence evaluation method combined with the single-stage gait sequence evaluation method.
The flow chart of the gait sequence evaluation method based on the quality dimension related to the single-stage gait sequence evaluation method in the embodiment of the invention is shown in fig. 1, and comprises the following steps:
and step S1, acquiring a single-stage gait sequence.
In the embodiment of the invention, the single-stage gait sequence is composed of a plurality of continuous single-frame gait images.
The gait sequence in the embodiment of the invention is required to have human shape characteristics, for example, the human shape segmentation result is not qualified if the human shape characteristics are not available due to the reasons that non-human figure input data is sent into the segmentation model, the quality of the human shape input data is poor, the accuracy of the segmentation model is insufficient, and the like, and needs to be collected again.
The gait sequence only keeps the shape of a walking person, and if the shape of a non-walking person such as squatting, riding, lying and the like is not qualified, the shape of the non-walking person needs to be collected again.
The gait sequence requires human body integrity, and if the conditions of upper half body loss, lower half body loss, left half body loss, right half body loss and the like exist, the loss exceeding 1/3 is regarded as unqualified and needs to be collected again.
In order to ensure that the acquired single-stage gait sequence meets the subsequent evaluation requirement, the embodiment of the invention further comprises the following steps after the single-stage gait sequence is acquired:
judging whether the single-stage gait sequence meets a preset pre-evaluation condition or not;
if yes, go to step S2;
and if not, deleting the single gait sequence.
The embodiment of the invention aims to control the acquisition quality from the data acquisition side so as to avoid the influence of the acquired data on subsequent gait recognition or gait retrieval by judging whether the single-stage gait sequence meets the preset pre-evaluation condition.
The pre-evaluation conditions in the embodiment of the invention comprise complete human shape, natural walking state and image quality. The reason for adopting the three pre-evaluation conditions is that there may be defects in the gait image collected by the collecting side camera, and the causes of the defects include three main categories, namely main characteristics, main behaviors and a collecting process. Wherein the defects caused by the main body characteristics comprise upper body loss, lower body loss, left body loss, right body loss, a backpack, a crutch, the color of clothes is consistent with the color of a background, and the shielding of other people/objects; defects caused by the main behaviors include stillness, riding, running, lifting legs, scratching the nose, playing a mobile phone, carrying criminals and the like; the defects caused by the acquisition process include complex, extremely strong or weak illumination of the field background, unfocused (low sharpness, fast motion leading to motion blur, etc.), and small image size or resolution.
The number and severity of defects will affect the performance of the gait recognition and gait retrieval system. Therefore, the embodiment of the invention selects the main body characteristics, the main body behaviors and the indexes in the acquisition process which can represent the causes of the defects to judge the quality of the single-stage gait sequence after the single-stage gait sequence is acquired, and deletes the acquired single-stage gait sequence when the quality of the single-stage gait sequence does not meet the pre-evaluation condition.
In the embodiment of the invention, whether a single-stage gait sequence meets a preset pre-evaluation condition is judged, and the method specifically comprises the following steps:
judging whether the gait images in the single-stage gait sequence contain complete humanoid figures or not;
judging whether the pedestrian in the single-stage gait sequence is in a natural walking state or not;
and judging whether the image quality of the gait image in the single-stage gait sequence is greater than a preset threshold value. The quality of the gait image comprises at least one of the resolution of the gait image, the size of the gait image and the frame number of the gait image; that is, it is necessary to determine whether the resolution of the gait image is greater than a preset resolution threshold, whether the size of the gait image is greater than a preset size threshold, whether the number of frames of the gait image is greater than a preset number of frames threshold, and the gait image can be entered into the gait database only if the number of frames exceeds the corresponding threshold. Wherein the image resolution is nominal using the number of image row and column pixels, and the height and width of the body in pixels can be used to measure the pixel range relative to the body features; the number of image frames is expressed by counting, and the gait image quality is influenced by over-small numerical value. And when the single-stage gait sequence has a complete human shape, pedestrians are in a natural walking state, and the quality of the gait image is greater than a preset threshold value, the single-stage gait sequence is considered to meet a preset pre-evaluation condition.
Further, the embodiment of the present invention determines whether the pedestrian in the single-stage gait sequence is in a natural walking state, and specifically includes:
respectively acquiring the mean value, standard deviation and covariance of two adjacent gait images in a single-stage gait sequence;
calculating first similarities of two adjacent gait images according to the mean value, the standard deviation and the covariance to obtain a plurality of first similarities;
and calculating the average value of the plurality of first similarity, and judging whether the average value is greater than a preset similarity threshold value, if so, determining that the pedestrian in the single-stage gait sequence is in a static state, and deleting the sequences in the static state which are unqualified.
Wherein, according to mean value, standard deviation and covariance, calculate the first similarity of two adjacent gait images, include specifically:
calculating a first similarity of two adjacent pedestrian images by using a formula (1):
in the formula (1), SSIM (K, L) is two adjacent gait patternsFirst degree of similarity, mu, like K and gait image LKAnd σKMean and standard deviation, mu, of the gait image KLAnd σLMean and standard deviation, σ, of the gait image L, respectivelyxyIs the covariance of the gait image K and the gait image L, c1And c2Is constant, set c1And c2The effect of (1) is to avoid instability caused when the denominator is close to 0.
And (3) traversing all gait images in the single-stage gait sequence by using the formula (1) to obtain a plurality of first similarities.
The SSIM in the embodiment of the invention is an index for measuring the similarity of pictures, and is a number from 0 to 1, and the larger the SSIM is, the smaller the difference between two images is. The method acquires the same area of two adjacent gait images, and then calculates the Similarity of the two areas through the SSIM (Structural Similarity) algorithm, thereby obtaining a plurality of first similarities.
And step S2, acquiring a first characteristic parameter of the single-stage gait sequence.
The first characteristic parameter in the embodiment of the present invention includes, but is not limited to, at least one of gait cycle quality, sharpness of human figure contour, illumination intensity and viewing angle.
Wherein, the illumination intensity is expressed by a quality score, thereby evaluating the intensity of the illumination intensity. The analysis object is a histogram corresponding to the unregulated pixel values of the whole image. The histogram when the lighting is normal is generally more extensive. When the illumination is too weak or too strong, the distribution of the gray values is concentrated at both ends of the histogram.
Suppose H0Is the histogram for standard lighting, and H is the histogram for the gait image to be evaluated. Can be according to H0And the difference in the gray value distribution of H to define the mass fraction.
Further, when the first characteristic parameter is the gait cycle quality, acquiring the first characteristic parameter of the single-stage gait sequence specifically includes:
and A, acquiring a gait silhouette sequence corresponding to the single-stage gait sequence.
In the embodiment of the invention, each gait image in the single-stage gait sequence is firstly segmented to obtain the gait silhouette image corresponding to each gait image, so that the gait silhouette sequence corresponding to the single-stage gait sequence can be further obtained.
After acquiring the gait silhouette sequence corresponding to the single-stage gait sequence, the method may further include:
normalizing and binarizing the gait silhouette sequence; specifically, the gait silhouette sequence is normalized, cut by a human center, and then binarized.
The non-normalized gait silhouette sequence is shown in fig. 2, and the normalized gait silhouette sequence is shown in fig. 3.
Step B, determining the gait cycle quality of the gait silhouette sequence, which specifically comprises the following steps:
step B1, calculating the contact ratio between the gait silhouette images in the gait silhouette sequence, which specifically comprises the following steps:
(b1) and acquiring a plurality of gait silhouette images with set frame number intervals in the gait silhouette sequence, wherein the set frame number intervals are determined by the size of the video capture frame rate.
(b2) Respectively calculating the contact ratio between two adjacent gait silhouette images in the plurality of gait silhouette images, and specifically comprising:
sequentially acquiring two adjacent gait silhouette images from the multiple gait silhouette images;
according to the pixel values of the same pixel positions of the two adjacent gait silhouette images, the contact ratio between the two adjacent gait silhouette images is calculated, and specifically, the contact ratio between the two adjacent gait silhouette images is calculated by using a formula (2).
In the formula (2), C is a coincidence degree, p and q are two adjacent gait silhouette images, g (m, n, p) is a pixel value of the gait silhouette image p at a corresponding pixel point (m, n), and g (m, n, q) is a pixel value of the gait silhouette image q at a corresponding pixel point (m, n), and the coincidence degree of the embodiment is obtained by traversing pixel values of all pixel points of the gait silhouette image.
And step B2, counting the number of the contact ratio larger than a preset contact ratio threshold value, and determining the gait cycle quality of the gait silhouette sequence according to the number.
The method for determining the gait cycle quality according to the quantity in the embodiment of the invention can be used for determining the gait cycle quality according to the preset mapping relation between the quantity and the gait cycle quality, for example: the gait cycle quality is 0.2 if the number of the contact ratio greater than the preset contact ratio threshold is 1, the gait cycle quality is 0.4 if the number of the contact ratio greater than the preset contact ratio threshold is 2, the gait cycle quality is 0.6 if the number of the contact ratio greater than the preset contact ratio threshold is 3, the gait cycle quality is 0.8 if the number of the contact ratio greater than the preset contact ratio threshold is 4, and the gait cycle quality is 1 if the number of the contact ratio greater than the preset contact ratio threshold is 5. The invention does not specifically limit the mapping relation between the quantity and the gait cycle quality, and can be set according to the actual situation.
In this embodiment, the gait is considered to be a recurrent gait if the coincidence degree is greater than the preset coincidence degree threshold, the number of recurrent gaits and the number of recurrent frames are counted, the gait cycle quality is evaluated, and the quality of the gait sequence is further obtained. A high quality gait sequence requires that the number of recurrent gaits be guaranteed. Namely, when the number of the repeated gaits is larger than the preset number threshold value, the gait sequence quality is considered to be better.
Further, when the first characteristic parameter is the definition of the contour, acquiring the first characteristic parameter of the single-stage gait sequence specifically includes:
step A, converting each gait image in the single-stage gait sequence into a gray image.
And B, filtering the gray level image in the single-stage gait sequence to obtain a filtering sequence.
In the embodiment of the invention, the Laplace operator is used for filtering the gray level image in the single-stage gait sequence, and the convolution kernel L of the used Laplace operator is as follows:
step C, calculating the variance of each filtering image in the filtering sequence, determining the definition of the human-shaped contour according to the variance, and specifically calculating the variance of each filtering image in the filtering sequence by using a formula (3):
in equation (3), LAP _ var (I) is the variance of the filtered image I, M and N are the width and height of the filtered image, respectively, M represents the M-th column of the image, N represents the N-th row of the image, L (M, N) represents the result of convolution at the filtered image I (M, N) with a convolution kernel L,
is the average absolute value, the calculation formula is as follows:
after the variance is obtained, the definition of the human-shaped contour is determined according to the variance, and specifically comprises the following steps: and comparing the variance with a preset variance threshold, wherein if the variance is greater than the preset variance threshold, the human figure contour is clear, the quality of the gait sequence is higher, and otherwise, the gait sequence is lower.
The variance used above may be the variance of a single filtered image, or may be the mean of the variances of all filtered images in the filtering sequence.
And step S3, evaluating the quality of the single gait sequence according to the first characteristic parameter to obtain the quality score of the single gait sequence.
When the first characteristic parameter is only one of the gait cycle quality, the definition of the human figure outline, the illumination intensity, the visual angle and the like, the mass fraction of the single-stage gait sequence is determined only according to the mapping relation between the pre-established first characteristic parameter and the mass fraction of the single-stage gait sequence.
When the first characteristic parameters are at least two of the gait cycle quality, the definition of the human figure contour, the illumination intensity, the visual angle and the like, regularizing each first characteristic parameter, specifically regularizing each first characteristic parameter according to a preset corresponding relation between each first characteristic parameter and a numerical value in a [0,1] interval or a numerical value interval. And then carrying out weighted average on the plurality of normalized first characteristic parameters to obtain the mass fraction of the single-stage gait sequence. The selection of the weights used in the weighted averaging needs to reflect the performance of the sample in the recognition environment.
The embodiment of the invention also comprises a multi-stage gait sequence evaluation method, and the evaluation method combines the mass fraction of the single-stage gait sequence to ensure that the obtained result is more accurate.
As shown in fig. 4, after step S3, the process of the embodiment of the present invention further includes:
step S4, acquiring a multi-segment gait sequence, which includes a plurality of single-segment gait sequences, and the premise of the multi-segment gait sequence is a plurality of single-segment gait sequences of the same target person.
And step S5, acquiring second characteristic parameters of the multiple gait sequences, wherein the second characteristic parameters include but are not limited to sequence number and view richness. The number of sequences refers to the number of single segment sequences. The view richness is the number of different views in a multi-segment gait sequence, and mainly represents the diversity of the views.
In the embodiment of the present invention, when the second characteristic parameter is view richness, acquiring the second characteristic parameter of the multi-segment gait sequence specifically includes:
step A, acquiring a walking track of a pedestrian in a multi-section gait sequence, and determining a plurality of characteristic points corresponding to the walking track by utilizing a track segmentation algorithm based on a minimum description length principle;
b, respectively connecting two adjacent characteristic points by using vectors to obtain a multi-section approximate track of the pedestrian;
and step C, determining the visual angle of each section of approximate track by combining the predetermined reference direction vector, and specifically comprising the following steps of: determining an included angle between the reference direction vector and each section of track vector in a clockwise direction by taking the reference direction vector as a start, wherein the included angle is a visual angle;
and D, counting the number of different visual angles, and recording the number of the different visual angles as the visual angle richness of the gait sequence.
Step S6, evaluating the quality of the multiple gait sequences according to the quality fraction of the single gait sequence and the second characteristic parameter to obtain the quality fraction of the multiple gait sequences, which specifically comprises the following steps:
and evaluating the quality of the multi-stage gait sequence by using a weighting evaluation model according to the quality fraction of the single-stage gait sequence and the second characteristic parameter to obtain the quality fraction of the multi-stage gait sequence.
Wherein the weighted average can be an arithmetic average or an exponential average.
When being an arithmetic mean, it is specifically:
QS＝α1(σ1M1+σ2M2+…+σQMQ)+α2(β1N1+…+βSNS) (7)
QS in the formula (7) is the mass fraction of the multi-stage gait sequence, MQIs the mass fraction, σ, of the Q-th single gait sequenceQIs the weight of the mass fraction of the Q-th single-step gait sequence, and σ1+σ2+…+σQ＝1；NSIs the S-th second characteristic parameter, and NSFor the normalized second feature parameter, when the second feature parameter only has the sequence number and the view richness, S is 2, βSIs the weight of the S-th second characteristic parameter, and1+β2+…+βQ＝1；α1and alpha2Is a weight, and α1+α21. All the weights need to be selected in consideration of the fact that the quality score of the gait image of the subject can reflect the performance of the sample in the recognition environment.
After the quality score is obtained, the embodiment of the invention can store and use the gait sequence in a grading way according to the quality score, and specifically comprises the following steps:
and step S7, comparing the mass fraction of the single-stage gait sequence or the mass fractions of the multiple-stage gait sequence with a corresponding preset fraction threshold value, and storing the single-stage gait sequence or the multiple-stage gait sequence in a gait base in a grading manner according to the comparison result.
According to the embodiment of the invention, the gait sequence can be divided into two stages according to the preset score threshold, wherein the low stage can be used for gait retrieval, and the high stage can be used for gait recognition.
A second aspect of the present invention provides a gait sequence evaluation system based on quality dimension, which uses the above gait sequence evaluation method based on quality dimension.
The gait sequence evaluation system based on the quality dimension has the structure shown in fig. 5, and comprises a sequence acquisition module 101, a feature acquisition module 102 and an evaluation module 103.
The sequence acquisition module 101 is used for acquiring a single-stage gait sequence;
the characteristic acquisition module 102 is configured to acquire a first characteristic parameter of a single gait sequence;
the evaluation module 103 is configured to evaluate the quality of the single gait sequence according to the first characteristic parameter to obtain a quality score of the single gait sequence.
The invention comprehensively considers the mass fractions of the local part (single-stage gait sequence in the multi-stage gait sequence) and the whole part (multi-stage gait sequence), and the obtained result has high precision.
Although the present application has been described with reference to a few embodiments, it should be understood that various changes, substitutions and alterations can be made herein without departing from the spirit and scope of the application as defined by the appended claims.
Claims (10)
1. A gait sequence assessment method based on quality dimension is characterized by comprising the following steps:
acquiring a single-stage gait sequence;
acquiring a first characteristic parameter of the single-stage gait sequence;
and evaluating the quality of the single-stage gait sequence according to the first characteristic parameter to obtain the quality fraction of the single-stage gait sequence.
2. The quality dimension-based gait sequence assessment method according to claim 1, characterized in that said first characteristic parameter comprises at least one of gait cycle quality, sharpness of humanoid contours, illumination intensity and perspective.
3. The quality dimension-based gait sequence assessment method according to claim 2, characterized in that when the first characteristic parameter is gait cycle quality, acquiring the first characteristic parameter of the single-stage gait sequence specifically comprises:
acquiring a gait silhouette sequence corresponding to the single-stage gait sequence;
determining the gait cycle quality of the gait silhouette sequence, which specifically comprises the following steps:
calculating the contact ratio between gait silhouette images in the gait silhouette sequence;
and counting the number of the contact ratios larger than a preset contact ratio threshold value, and determining the gait cycle quality of the gait silhouette sequence according to the number.
4. The gait sequence evaluation method based on quality dimension according to claim 3, characterized in that the calculating of the degree of coincidence between the gait silhouette images in the gait silhouette sequence specifically comprises:
acquiring a plurality of gait silhouette images with set frame number intervals in a gait silhouette sequence;
respectively calculating the contact ratio between two adjacent gait silhouette images in the plurality of gait silhouette images, which specifically comprises the following steps:
sequentially acquiring two adjacent gait silhouette images from the multiple gait silhouette images;
and calculating the contact ratio between the two adjacent gait silhouette images according to the pixel values at the same pixel positions of the two adjacent gait silhouette images.
5. The gait sequence evaluation method based on quality dimension according to claim 2, wherein when the first characteristic parameter is the definition of human figure, the acquiring the first characteristic parameter of the single-stage gait sequence specifically comprises:
converting each gait image in the single gait sequence into a gray image;
filtering the gray level image to obtain a filtering sequence;
and calculating the variance of each filtering image in the filtering sequence, and determining the definition of the human-shaped contour according to the variance.
6. The quality dimension based gait sequence assessment method according to claim 1, characterized by, after acquiring a single-stage gait sequence, further comprising:
judging whether the single gait sequence meets a preset pre-evaluation condition or not; if yes, acquiring a first characteristic parameter of the single-stage gait sequence;
if not, deleting the single gait sequence;
wherein the pre-evaluation conditions comprise complete human shape, natural walking state and image quality;
the image quality comprises at least one of the resolution of the gait image, the size of the gait image and the frame number of the gait image;
correspondingly, whether the single-stage gait sequence meets a preset pre-evaluation condition is judged, and the method specifically comprises the following steps:
judging whether the gait images in the single-stage gait sequence contain complete humanoid figures or not;
judging whether the pedestrian in the single-stage gait sequence is in a natural walking state or not;
and judging whether the image quality of the gait image in the single-stage gait sequence is greater than a preset quality threshold value or not.
7. The gait sequence assessment method according to claim 6, characterized in that the step of determining whether the pedestrian in the single-stage gait sequence is in a natural walking state comprises:
respectively acquiring the mean value, the standard deviation and the covariance of two adjacent gait images in the single-stage gait sequence;
calculating first similarities of two adjacent gait images according to the mean value, the standard deviation and the covariance to obtain a plurality of first similarities;
and calculating the average value of the plurality of first similarity, and judging whether the average value is greater than a preset similarity threshold value, if so, determining that the pedestrian in the single-stage gait sequence is in a static state.
8. The quality dimension based gait sequence assessment method according to any one of claims 1 to 7, characterized by further comprising:
acquiring a multi-segment gait sequence, wherein the multi-segment gait sequence comprises a plurality of single-segment gait sequences;
acquiring second characteristic parameters of the multi-segment gait sequence;
evaluating the quality of the multi-segment gait sequence according to the quality fraction of the single-segment gait sequence and the second characteristic parameter to obtain the quality fraction of the multi-segment gait sequence;
the second characteristic parameters include the number of sequences and the view richness.
9. The quality dimension based gait sequence assessment method according to claim 8, characterized by further comprising:
and comparing the mass fraction of the single gait sequence or the mass fractions of the multiple gait sequences with a corresponding preset fraction threshold, and storing the single gait sequence or the multiple gait sequences in a gait base database in a grading manner according to a comparison result.
10. A gait sequence assessment system based on quality dimensions, comprising:
the sequence acquisition module is used for acquiring a single-stage gait sequence;
the characteristic acquisition module is used for acquiring a first characteristic parameter of the single gait sequence;
and the evaluation module is used for evaluating the quality of the single gait sequence according to the first characteristic parameter to obtain the quality score of the single gait sequence.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210249668.4A CN114612934A (en) | 2022-03-14 | 2022-03-14 | Gait sequence evaluation method and system based on quality dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210249668.4A CN114612934A (en) | 2022-03-14 | 2022-03-14 | Gait sequence evaluation method and system based on quality dimension |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114612934A true CN114612934A (en) | 2022-06-10 |
Family
ID=81863875
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210249668.4A Pending CN114612934A (en) | 2022-03-14 | 2022-03-14 | Gait sequence evaluation method and system based on quality dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114612934A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116524592A (en) * | 2023-04-18 | 2023-08-01 | 凯通科技股份有限公司 | Gait sequence silhouette generation method and device, electronic equipment and storage medium |
-
2022
- 2022-03-14 CN CN202210249668.4A patent/CN114612934A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116524592A (en) * | 2023-04-18 | 2023-08-01 | 凯通科技股份有限公司 | Gait sequence silhouette generation method and device, electronic equipment and storage medium |
| CN116524592B (en) * | 2023-04-18 | 2024-02-06 | 凯通科技股份有限公司 | Gait sequence silhouette generation method and device, electronic equipment and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109857889B (en) | Image retrieval method, device and equipment and readable storage medium | |
| CN111104867B (en) | Recognition model training and vehicle re-recognition method and device based on part segmentation | |
| US9367730B2 (en) | Method and system for automated face detection and recognition | |
| CN113065558A (en) | Lightweight small target detection method combined with attention mechanism | |
| CN109145708B (en) | Pedestrian flow statistical method based on RGB and D information fusion | |
| CN107622229A (en) | A kind of video frequency vehicle based on fusion feature recognition methods and system again | |
| CN102422325B (en) | Pattern recognition apparatus and method therefor configured to recognize object and another lower-order object | |
| CN110909605A (en) | Cross-modal pedestrian re-identification method based on contrast correlation | |
| CN109271932A (en) | Pedestrian based on color-match recognition methods again | |
| CN111814690B (en) | Target re-identification method, device and computer readable storage medium | |
| CN111046789A (en) | Pedestrian re-identification method | |
| US8094971B2 (en) | Method and system for automatically determining the orientation of a digital image | |
| CN104463240B (en) | A kind of instrument localization method and device | |
| CN110909565B (en) | Image recognition and pedestrian re-recognition method and device, electronic and storage equipment | |
| CN115131760B (en) | Lightweight vehicle tracking method based on improved feature matching strategy | |
| CN112287802A (en) | Face image detection method, system, storage medium and equipment | |
| CN111860196B (en) | Hand operation action scoring device, method and computer readable storage medium | |
| CN112101195A (en) | Crowd density estimation method and device, computer equipment and storage medium | |
| CN112287906A (en) | Template matching tracking method and system based on depth feature fusion | |
| CN114612934A (en) | Gait sequence evaluation method and system based on quality dimension | |
| CN102148919B (en) | Method and system for detecting balls | |
| CN107145820B (en) | Binocular positioning method based on HOG characteristics and FAST algorithm | |
| Patil et al. | Survey on approaches used for image quality assessment | |
| CN114998980B (en) | Iris detection method and device, electronic equipment and storage medium | |
| CN115601687A (en) | Intelligent processing method for on-site survey data in insurance claim settlement process |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |