A control evaluation method and system for gait recognition and collection devices

Abstract

The present invention discloses a method and system for evaluating the deployment of a gait recognition and acquisition device. The method includes obtaining a video of a pedestrian collected by the acquisition device within a preset time period; extracting the pedestrian's gait sequence based on the pedestrian video; determining the missed detection rate and gait sequence efficiency of the acquisition device based on the gait sequence; and evaluating the deployment quality of the acquisition device based on the missed detection rate and gait sequence efficiency. The method and system for evaluating the deployment of a gait recognition and acquisition device of the present invention utilize the missed detection rate and gait sequence efficiency to evaluate the deployment quality of the acquisition device, and can obtain accurate and effective deployment quality of the acquisition device. The method of the present invention can improve the quality of video data collected by the acquisition device, thereby reducing the difficulty of gait recognition in the security process, improving recognition speed and recognition accuracy, and providing a basis for the selection of gait capture points.

Description

Technical Field

[0001] The invention discloses a control and assessment method and system for a gait recognition and collection device, belonging to the technical field of security video processing. Background Art

[0002] Video surveillance has become a crucial technology in the security field. To ensure the effectiveness of video information, the video performance of security systems, centered around acquisition devices (such as cameras), must meet specific requirements. Therefore, video performance evaluation is essential.

[0003] Existing video performance assessments are primarily based on image quality, such as using the signal-to-noise ratio (SNR) and AI algorithms that incorporate human visual characteristics to assess image clarity. However, in gait recognition systems, conditions such as leg occlusion and short walking times that affect gait feature extraction are often unrelated to image clarity. On the other hand, gait recognition systems can still extract gait features from videos that are blurry, have poor clarity, or exhibit color variations. Therefore, the quality of videos used for gait recognition cannot be fully evaluated using traditional security video quality assessment standards.

[0004] Existing techniques for evaluating video performance rely solely on the image itself, without considering the impact of the video capture device on the evaluation results. For example, if the capture device is poorly positioned, resulting in poor quality video images, gait recognition using the captured data will not achieve ideal results, increasing recognition difficulty and slowing down the process. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for evaluating the placement of a gait recognition acquisition device, so as to solve the technical problems in the prior art that the acquisition device is poorly placed, resulting in poor quality of the video images it collects, increased difficulty in gait recognition, and reduced speed of gait recognition.

[0006] A first aspect of the present invention provides a method for evaluating the deployment of a gait recognition and acquisition device, comprising:

[0007] Obtaining pedestrian videos collected by a collection device within a preset time period;

[0008] extracting a pedestrian's gait sequence based on the pedestrian video;

[0009] Determining a missed beat rate of a collection device and a gait sequence efficiency rate according to the gait sequence;

[0010] The control quality of the acquisition device is evaluated according to the missed beat rate and the gait sequence efficiency.

[0011] Preferably, determining the missed beat rate of the acquisition device according to the gait sequence specifically includes:

[0012] The missed shot rate is determined according to the total number of pedestrians in the pedestrian video and the number of pedestrians corresponding to the gait sequence.

[0013] Preferably, determining the gait sequence efficiency according to the gait sequence specifically includes:

[0014] A quality assessment is performed on the gait sequence, and a gait sequence efficiency is determined based on the gait sequence meeting a preset quality.

[0015] Preferably, performing a quality assessment on the gait sequence and determining the gait sequence efficiency based on the gait sequence meeting the preset quality specifically includes:

[0016] performing a quality assessment on the gait sequence;

[0017] The gait sequence efficiency is determined according to the number of gait sequences that meet the preset quality and the total number of gait sequences.

[0018] Preferably, evaluating the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency specifically includes:

[0019] Obtaining weights of the missed beat rate and the gait sequence efficiency;

[0020] The control quality of the acquisition device is evaluated according to the missed beat rate, the gait sequence efficiency and the corresponding weights.

[0021] Preferably, after determining the missed beat rate of the acquisition device and the gait sequence efficiency according to the gait sequence, the method further includes:

[0022] Extract facial features from gait sequences that meet preset quality;

[0023] The facial image extraction rate is determined according to the number of gait sequences corresponding to the extracted facial features.

[0024] Preferably, the facial image extraction rate is determined according to the number of gait sequences corresponding to the extracted facial features, specifically including:

[0025] determining a facial image extraction rate based on the number of gait sequences corresponding to the extracted facial features and the number of gait sequences meeting a preset quality;

[0026] Accordingly, evaluating the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency specifically includes:

[0027] The control quality of the acquisition device is evaluated according to the missed shot rate, the gait sequence efficiency rate and the face image extraction rate.

[0028] Preferably, evaluating the control quality of the acquisition device according to the missed shot rate, the gait sequence efficiency rate, and the face image extraction rate specifically includes:

[0029] Obtaining weights of the missed beat rate, the gait sequence efficiency, and the face image extraction rate;

[0030] The control quality of the acquisition device is evaluated according to the missed shot rate, the gait sequence efficiency, the face image extraction rate and the corresponding weights.

[0031] A first aspect of the present invention provides a deployment control and evaluation system for a gait recognition and collection device, comprising:

[0032] A video acquisition module, which is used to acquire pedestrian videos collected by a collection device within a preset time period;

[0033] a gait sequence extraction module, the gait sequence extraction module being used to extract the pedestrian's gait sequence based on the pedestrian video;

[0034] a first parameter determination module, wherein the first parameter acquisition module is used to determine a missed beat rate and a gait sequence efficiency of a collection device according to the gait sequence;

[0035] An evaluation module is used to evaluate the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency.

[0036] Compared with the prior art, the control and assessment method and system for gait recognition and collection devices of the present invention have the following beneficial effects:

[0037] The control evaluation method and system for gait recognition and acquisition devices of the present invention use the missed detection rate and gait sequence efficiency to evaluate the control quality of the acquisition device, and can obtain accurate and effective control quality of the acquisition device. Based on the results obtained, the staff can judge whether the control position of the acquisition device is appropriate. If it is not appropriate, adjustments can be made, and then the control evaluation can be performed again using this method until a satisfactory result that meets the preset conditions (such as thresholds) is obtained. The method of the present invention can improve the quality of video data collected by the acquisition device, thereby reducing the difficulty of gait recognition in the security process, and improving the recognition speed and recognition accuracy. The present invention provides a basis for the selection of gait capture points, improves the efficiency of fault location and problem handling, and supports automated fault analysis and problem handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 A flowchart of a method for evaluating the deployment of a gait recognition and acquisition device provided by an embodiment of the present invention;

[0039] Figure 2A schematic diagram of the structure of a control and evaluation system for a gait recognition and collection device provided in an embodiment of the present invention.

[0040] In the figure, 101 is a video acquisition module; 102 is a gait sequence extraction module; 103 is a parameter determination module; and 104 is an evaluation module. DETAILED DESCRIPTION

[0041] In the following description, specific details such as particular system structures and techniques are provided for purposes of illustration, not limitation, to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, 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.

[0042] The process of the control evaluation method for the gait recognition and collection device according to the embodiment of the present invention is as follows: Figure 1 Shown, including:

[0043] Step 1: Obtain pedestrian videos collected by a collection device within a preset time period.

[0044] The preset time period in this embodiment may be 3 hours, 5 hours, 10 hours, 24 hours, 2 days, etc., and the acquisition device may be a camera, which is not limited in the present invention.

[0045] Step 2: Extract the pedestrian gait sequence based on the pedestrian video.

[0046] The gait sequence in this embodiment is composed of multiple continuous single-frame gait images.

[0047] Step 3: Determine the missed beat rate of the acquisition device and the gait sequence efficiency according to the gait sequence.

[0048] In this embodiment, the missed shot rate is determined based on the total number of pedestrians in the pedestrian video and the number of pedestrians whose gait sequences are extracted. For example, the missed shot rate can be determined according to formula (1):

[0049]

[0050] Where FPR is the missed shot rate, NP is the total number of pedestrians, and NG is the number of pedestrians whose gait sequences are extracted.

[0051] The method for obtaining the total number of pedestrians and extracting pedestrian gait sequences may use image segmentation methods in the prior art, such as texture-based image segmentation methods, threshold-based image segmentation methods, etc., and the present invention does not impose any restrictions on this.

[0052] Furthermore, the process of determining the gait sequence efficiency in this embodiment is as follows:

[0053] The quality of the gait sequence is evaluated, and the gait sequence efficiency is determined according to the number of gait sequences that meet the preset quality and the total number of gait sequences. For example, the gait sequence efficiency can be determined according to formula (2):

[0054]

[0055] Where GEFF is the gait sequence efficiency, EG is the number of gait sequences that meet the preset quality, and WG is the total number of gait sequences.

[0056] The method for evaluating the quality of gait sequences can use existing quality evaluation methods, such as:

[0057] A first characteristic parameter of a pedestrian's gait sequence is obtained, and the quality of the gait sequence is evaluated according to the first characteristic parameter to obtain a quality score of the gait sequence.

[0058] The first characteristic parameter includes but is not limited to at least one of gait cycle quality, clarity of human figure outline, lighting intensity and viewing angle.

[0059] The lighting intensity is expressed as a quality score, which can be used to evaluate the strength of the lighting intensity. The analysis object is the histogram corresponding to the unnormalized pixel values ​​of the entire image. Figure 1 When the lighting is too weak or too strong, the distribution of grayscale values ​​is concentrated at both ends of the histogram.

[0060] When the first characteristic parameter is the gait cycle quality, obtaining the first characteristic parameter of the gait sequence specifically includes:

[0061] Step A: Obtain a gait silhouette sequence corresponding to the gait sequence.

[0062] In the embodiment of the present invention, each gait image in the gait sequence is first segmented to obtain a gait silhouette image corresponding to each gait image, and then a gait silhouette sequence corresponding to the gait sequence can be obtained.

[0063] After obtaining the gait silhouette sequence corresponding to the gait sequence, the following steps may also be performed:

[0064] The gait silhouette sequence is normalized and binarized; specifically, the gait silhouette sequence is normalized, cropped with the person as the center, and then the normalized gait silhouette sequence is binarized.

[0065] Step B: determining the gait cycle quality of the gait silhouette sequence, specifically comprising:

[0066] Step B1: calculating the overlap between gait silhouette images in the gait silhouette sequence, specifically including:

[0067] (b1) obtaining a plurality of gait silhouette images with a set frame interval in the gait silhouette sequence, wherein the set frame interval is determined by the frame rate of the video capture.

[0068] (b2) respectively calculating the overlap between two adjacent gait silhouette images in the plurality of gait silhouette images, specifically including:

[0069] sequentially acquiring two adjacent gait silhouette images from a plurality of gait silhouette images;

[0070] According to the pixel values ​​at the same pixel position of the two adjacent gait silhouette images, the overlap between the two adjacent gait silhouette images is calculated. Specifically, the overlap between the two adjacent gait silhouette images is calculated using formula (3).

[0071]

[0072] In formula (3), C is the overlap, p and q are two adjacent gait silhouette images, g(m,n,p) is the pixel value of the gait silhouette image p at its corresponding pixel point (m,n), and g(m,n,q) is the pixel value of the gait silhouette image q at its corresponding pixel point (m,n). The overlap in this embodiment is obtained by traversing the pixel values ​​at all pixel points of the gait silhouette image.

[0073] Step B2: Count the number of times the overlap is greater than a preset overlap threshold, and determine the gait cycle quality of the gait silhouette sequence according to the number.

[0074] In an embodiment of the present invention, the method for determining gait cycle quality based on quantity may be to determine gait cycle quality based on a pre-set mapping relationship between quantity and gait cycle quality. For example, if the number of instances where the overlap exceeds a pre-set overlap threshold is 1, the gait cycle quality is 0.2; if the number of instances where the overlap exceeds the pre-set overlap threshold is 2, the gait cycle quality is 0.4; if the number of instances where the overlap exceeds the pre-set overlap threshold is 3, the gait cycle quality is 0.6; if the number of instances where the overlap exceeds the pre-set overlap threshold is 4, the gait cycle quality is 0.8; and if the number of instances where the overlap exceeds the pre-set overlap threshold is 5, the gait cycle quality is 1. The present invention does not impose any specific limitation on the mapping relationship between quantity and gait cycle quality, and the mapping relationship may be set according to actual circumstances.

[0075] In this embodiment, a gait is considered recurring if the overlap exceeds a preset overlap threshold. The number of recurring gaits and the number of recurring frames are counted to evaluate the gait cycle quality and further determine the quality of the gait sequence. A high-quality gait sequence requires a sufficient number of recurring gaits. Specifically, when the number of recurring gaits exceeds a preset threshold, the gait sequence is considered of good quality.

[0076] Furthermore, when the first characteristic parameter is the clarity of the human figure outline, obtaining the first characteristic parameter of the gait sequence specifically includes:

[0077] Step A: Convert each gait image in the gait sequence into a grayscale image.

[0078] Step B: Filter the grayscale image in the gait sequence to obtain a filtered sequence.

[0079] In the embodiment of the present invention, the grayscale image in the gait sequence is filtered using the Laplacian operator, and the convolution kernel L of the Laplacian operator used is as follows:

[0080]

[0081] Step C: Calculate the variance of each filtered image in the filtering sequence, and determine the clarity of the human figure outline based on the variance. Specifically, use formula (4) to calculate the variance of each filtered image in the filtering sequence:

[0082]

[0083] In formula (4), 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 mth column of the image, n represents the nth row of the image, L(m,n) represents the result of convolution of the filtered image I(m,n) with the convolution kernel L, is the mean absolute value, which is calculated as follows:

[0084]

[0085] After obtaining the variance, the clarity of the human figure outline is determined according to the variance as follows: the variance is compared with a preset variance threshold. If the variance is greater than the preset variance threshold, the human figure outline is clear and the quality of the gait sequence is high, otherwise it is low.

[0086] The variance used above may be the variance of a single filtered image or the mean of the variances of all filtered images in the filtering sequence.

[0087] The quality of the gait sequence is then evaluated based on the first characteristic parameter to obtain a quality score of the gait sequence. When the first characteristic parameter is only one of the gait cycle quality, clarity of the human outline, lighting intensity, and viewing angle, the quality score of the gait sequence only needs to be determined according to the pre-established mapping relationship between the first characteristic parameter and the quality score of the gait sequence.

[0088] When the first feature parameters are at least two of the following: gait cycle quality, human figure outline clarity, lighting intensity, and viewing angle, each first feature parameter is regularized. Specifically, regularization can be performed based on a pre-set correspondence between each first feature parameter and a value or numerical range in the interval [0, 1]. A weighted average of the regularized first feature parameters is then performed to obtain a quality score for the gait sequence. The weights used in the weighted average should reflect the performance of the sample in the recognition environment.

[0089] In the embodiment of the present invention, a gait sequence meeting a preset quality may be a gait sequence whose quality score determined by the above method is greater than or equal to a preset threshold.

[0090] Step 4: Evaluate the quality of the collection device deployment based on missed shot rate and gait sequence efficiency, including:

[0091] Step 41: Obtain the weights of the missed beat rate and the gait sequence efficiency.

[0092] The weights can be set according to the importance of each indicator.

[0093] Step 42: Evaluate the control quality of the acquisition device based on the missed shot rate, gait sequence efficiency, and corresponding weights. For example, the control quality of the acquisition device can be evaluated according to formula (6):

[0094] SCORE=FPR×p1+GEFF×p2 (6)

[0095] Where SCORE is the control quality score of the acquisition device, FPR is the missed shot rate, GEFF is the gait sequence efficiency, and p1 and p2 are weights, which can be set according to the importance of each indicator.

[0096] To further improve the accuracy of the collection device deployment assessment, the present invention, after determining the missed capture rate and gait sequence efficiency of the collection device based on the gait sequence, further includes:

[0097] Extract facial features from gait sequences that meet preset quality;

[0098] The facial image extraction rate is determined based on the number of gait sequences corresponding to the extracted facial features and the number of gait sequences that meet the preset quality. For example, the facial image extraction rate can be determined according to formula (7):

[0099]

[0100] Where FEFF is the face image extraction rate, EG is the number of gait sequences that meet the preset quality, and EGF is the number of gait sequences corresponding to the extracted facial features.

[0101] In this embodiment, the method used to extract facial features from gait sequences that meet the preset quality may be a convolutional neural network method or a Histograms of Oriented Gradients (HOG method for short).

[0102] Accordingly, the control quality of the acquisition device can be evaluated based on the missed shot rate, gait sequence efficiency, and face image extraction rate. Specifically:

[0103] Obtain the weights of missed shot rate, gait sequence efficiency and face image extraction rate. The weights can be determined according to the importance of each indicator.

[0104] The control quality of the acquisition device is evaluated based on the missed shot rate, gait sequence efficiency, face image extraction rate and corresponding weights. For example, the control quality of the acquisition device can be evaluated according to formula (8):

[0105] SCORE=FPR×p1+GEFF×p2+FEFF×p3 (8)

[0106] In the formula, SCORE is the control quality score of the acquisition device, FPR is the missed capture rate, GEFF is the gait sequence efficiency, FEFF is the face image extraction rate, and p1, p2, and p3 are weights. The weights can be set based on the importance of each indicator. For example, using a percentage system, the score weights p1, p2, and p3 can be set to 50, 40, and 10, respectively.

[0107] The second aspect of the present invention provides a control and evaluation system for a gait recognition and collection device, such as Figure 2 As shown, it includes a video acquisition module 101, a gait sequence extraction module 102, a parameter determination module 103 and an evaluation module 104.

[0108] The video acquisition module is used to acquire pedestrian videos collected by the acquisition device within a preset time period;

[0109] The gait sequence extraction module is used to extract the pedestrian's gait sequence based on the pedestrian video;

[0110] The parameter acquisition module is used to determine the missed beat rate of the acquisition device and the gait sequence efficiency according to the gait sequence;

[0111] The evaluation module is used to evaluate the deployment quality of the acquisition device based on the missed shot rate and gait sequence efficiency.

[0112] The control evaluation method and system for gait recognition and acquisition devices of the present invention use the missed detection rate and gait sequence efficiency to evaluate the control quality of the acquisition device, and can obtain accurate and effective control quality of the acquisition device. Based on the results obtained, the staff can judge whether the control position of the acquisition device is appropriate. If it is not appropriate, adjustments can be made, and then the control evaluation can be performed again using this method until a satisfactory result that meets the preset conditions (such as thresholds) is obtained. The method of the present invention can improve the quality of video data collected by the acquisition device, thereby reducing the difficulty of gait recognition in the security process, and improving the recognition speed and recognition accuracy. The present invention provides a basis for the selection of gait capture points, improves the efficiency of fault location and problem handling, and supports automated fault analysis and problem handling.

[0113] The above descriptions are merely a few embodiments of the present application and do not constitute any form of limitation to the present application. Although the present application discloses the preferred embodiments as above, they are not intended to limit the present application. Any technical personnel familiar with the present profession, without departing from the scope of the technical solution of the present application, using the technical content disclosed above to make slight changes or modifications are equivalent to equivalent implementation cases and fall within the scope of the technical solution.

Claims

1. A method for evaluating the deployment of a gait recognition and acquisition device, characterized in that: include: Obtaining pedestrian videos collected by a collection device within a preset time period; extracting a pedestrian's gait sequence based on the pedestrian video; Determining a missed beat rate of a collection device and a gait sequence efficiency rate according to the gait sequence; evaluating the control quality of the acquisition device according to the missed beat rate and the gait sequence efficiency; Determining a missed shot rate of a collection device according to the gait sequence specifically includes: determining the missed shot rate according to the total number of pedestrians in the pedestrian video and the number of pedestrians corresponding to the gait sequence; Determining the gait sequence efficiency according to the gait sequence specifically includes: performing a quality assessment on the gait sequence, and determining the gait sequence efficiency according to the gait sequence meeting a preset quality, specifically including: performing a quality assessment on the gait sequence; The gait sequence efficiency is determined according to the number of gait sequences that meet the preset quality and the total number of gait sequences.

2. The method for evaluating the deployment of a gait recognition and collection device according to claim 1, wherein: Evaluating the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency specifically includes: Obtaining weights of the missed beat rate and the gait sequence efficiency; The control quality of the acquisition device is evaluated according to the missed beat rate, the gait sequence efficiency and the corresponding weights.

3. The method for evaluating the deployment of a gait recognition and collection device according to claim 1, wherein: After determining the missed beat rate of the acquisition device and the gait sequence efficiency according to the gait sequence, the method further includes: Extract facial features from gait sequences that meet preset quality; The facial image extraction rate is determined according to the number of gait sequences corresponding to the extracted facial features.

4. The method for evaluating the deployment of a gait recognition and collection device according to claim 3, wherein: The facial image extraction rate is determined based on the number of gait sequences corresponding to the extracted facial features, specifically including: determining a facial image extraction rate based on the number of gait sequences corresponding to the extracted facial features and the number of gait sequences meeting a preset quality; Accordingly, evaluating the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency specifically includes: The control quality of the acquisition device is evaluated according to the missed shot rate, the gait sequence efficiency rate and the face image extraction rate.

5. The method for evaluating the deployment of a gait recognition and collection device according to claim 4, characterized in that: The control quality of the acquisition device is evaluated according to the missed shot rate, the gait sequence efficiency rate, and the face image extraction rate, specifically including: Obtaining weights of the missed beat rate, the gait sequence efficiency, and the face image extraction rate; The control quality of the acquisition device is evaluated according to the missed shot rate, the gait sequence efficiency, the face image extraction rate and the corresponding weights.

6. A control and assessment system for a gait recognition and collection device based on the control and assessment method for a gait recognition and collection device according to any one of claims 1 to 5, characterized in that: include: A video acquisition module, which is used to acquire pedestrian videos collected by a collection device within a preset time period; a gait sequence extraction module, the gait sequence extraction module being used to extract the pedestrian's gait sequence based on the pedestrian video; a parameter determination module, wherein the parameter acquisition module is used to determine the missed beat rate of the acquisition device and the gait sequence efficiency according to the gait sequence; An evaluation module is used to evaluate the control quality of the acquisition device according to the missed shot rate and the gait sequence efficiency.

Images