A target person processing method and device, a storage medium, and a server

Abstract

Embodiments of the present application disclose a target person processing method and device, a storage medium and a server. The method comprises: acquiring a shooting video, and screening a candidate face image from the shooting video; matching the candidate face image with at least one target face image; if there is a matching face image matched with the candidate face image in the target face image, acquiring grade information and a code of the matching person corresponding to the matching face image; determining a target decision tree corresponding to the grade information from a plurality of decision trees according to the grade information, the decision tree comprising a plurality of decision nodes with hierarchical relationships, each decision node corresponding to a sub-processing mode; determining a target sub-processing mode from a plurality of sub-processing modes based on the code, and obtaining a processing decision of the matching person. The dangerous person is automatically identified by the face recognition mode, and the coping decision for the dangerous person is decided by the decision tree, which can improve the intelligent degree and the supervision accuracy of the monitoring.

Description

Technical Field

[0001] This application relates to the field of computers, and specifically to a method, apparatus, computer-readable storage medium, and server for processing a target person. Background Technology

[0002] In recent years, with the development of the times, more and more enterprises have emerged. In order to avoid dangerous personnel from entering the enterprise and causing unnecessary trouble, enterprises will set up monitoring rooms to monitor the entry of dangerous personnel through personnel monitoring.

[0003] In existing technologies, personnel monitoring is carried out by recording with video equipment. Typically, multiple staff members observe on different monitoring screens. When a dangerous person is observed, human feedback is provided, and then a response strategy for dealing with the dangerous person is decided, and temporary personnel deployment is carried out.

[0004] In the process of researching and practicing existing technologies, the inventors of this application found that the entire monitoring and management process in the existing technologies is rather rushed, which is not conducive to timely response by personnel. At the same time, monitoring by having multiple staff members observe the surveillance camera footage is inconvenient, wastes manpower, and has poor monitoring accuracy. The entire monitoring and management process has a poor level of intelligence, requires multiple people for macro-control, wastes manpower, and has poor monitoring effect. Summary of the Invention

[0005] This application provides a method and apparatus for processing target individuals, which can avoid human monitoring and improve the intelligence level and accuracy of monitoring.

[0006] To address the aforementioned technical problems, this application provides the following technical solutions:

[0007] A method for handling a target person includes:

[0008] Acquire a video recording, and filter out candidate face images from the video recording. The candidate face images are images that possess the facial features of the candidate.

[0009] The candidate face image is matched with at least one target face image;

[0010] If a matching face image exists in the target face image that matches the candidate face image, obtain the level information and code information of the matching person corresponding to the matching face image;

[0011] Based on the level information, a target decision tree corresponding to the level information is determined from multiple decision trees. The target decision tree includes multiple decision nodes with hierarchical relationships, and each decision node corresponds to a sub-processing method.

[0012] Based on the code information, the target sub-processing method is determined from the multiple sub-processing methods to obtain the processing decision for the matched person.

[0013] A device for processing a target person, comprising:

[0014] The filtering module is used to acquire the captured video and filter out candidate face images from the captured video. The candidate face images are images that have the facial features of the candidate.

[0015] A matching module is used to match the candidate face image with at least one target face image;

[0016] The acquisition module is used to acquire the level information and code information of the matching person corresponding to the matching face image if there is a matching face image in the target face image that matches the candidate face image;

[0017] The first determining module is used to determine the target decision tree corresponding to the level information from multiple decision trees based on the level information. The target decision tree includes multiple decision nodes with hierarchical relationships, and each decision node corresponds to a sub-processing method.

[0018] The second determining module is used to determine the target sub-processing method from the multiple sub-processing methods based on the code information, and to obtain the processing decision of the matched person.

[0019] In some embodiments, the code information includes at least one sub-code information segment, each sub-code information segment corresponding to a decision node level, and each decision node in the decision node level is assigned a code range. The second determining module includes:

[0020] The first determining submodule is used to determine the target subprocessing method of the matched person at each decision node level based on the code range corresponding to the decision node in each decision node level and each segment of sub-code information in the code information.

[0021] The second determining submodule is used to determine the processing method of the matched person based on the target subprocessing method.

[0022] In some embodiments, the first determining submodule includes:

[0023] The acquisition unit is used to acquire the target sub-codes included in the sub-code information in the code information, and the code range of each decision node in the decision node hierarchy corresponding to the sub-code information.

[0024] The first determining unit is used to determine the decision node corresponding to the code range including the target sub-code as the target decision node, and to determine the sub-processing method corresponding to the target decision node as the target sub-processing method, so as to obtain the target sub-processing method of the matched person at each decision node level.

[0025] In some embodiments, the apparatus further includes:

[0026] An extraction module is used to extract at least two types of facial feature vectors from the candidate face image to obtain a candidate facial feature group of the candidate face image;

[0027] The matching module includes:

[0028] The matching submodule is used to match the image similarity between the candidate face image and at least one target face image in the target face image group;

[0029] The third determination submodule is used to determine target face images with image similarity greater than a preset similarity threshold as candidate matching face images;

[0030] The fourth determination submodule is used to obtain the candidate matching facial feature group of the candidate matching face image. If the candidate facial feature group is the same as the candidate matching facial feature group, then the candidate matching face image is determined as the matching face image.

[0031] In some embodiments, the extraction module includes:

[0032] The acquisition submodule is used to acquire at least two facial features from the candidate face image;

[0033] The segmentation submodule is used to segment the image for each facial feature, resulting in multiple segmented images;

[0034] The binary processing submodule is used to perform binary processing on the pixel values ​​of each pixel in the segmented image to obtain the segmented image after binary processing.

[0035] The calculation submodule is used to calculate the feature vector of the processed segmented image to obtain at least two kinds of facial feature vectors.

[0036] In some embodiments, the binary processing submodule includes:

[0037] The calculation unit is used to obtain the pixel value of the center pixel of the segmented image and calculate the pixel difference between each other pixel in the segmented image and the center pixel.

[0038] The second determining unit is used to determine pixels whose pixel differences are within a preset pixel difference range as 1, and pixels whose pixel differences are not within the preset pixel difference range as 0, to obtain a segmented image after binary processing.

[0039] In some embodiments, the apparatus further includes:

[0040] The preprocessing module is used to preprocess the candidate face image to obtain a preprocessed candidate face image;

[0041] The correction module is used to correct the preprocessed candidate face image using a preset correction model to obtain the corrected candidate face image.

[0042] In some embodiments, the preprocessing module includes:

[0043] The grayscale processing and edge detection processing submodule is used to perform grayscale processing and edge detection processing on the candidate face image to obtain a preprocessed candidate face image.

[0044] A computer-readable storage medium storing a plurality of instructions adapted for loading by a processor to perform the steps in the aforementioned method for processing a target person.

[0045] A server includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the steps in the method for processing a target person as described above.

[0046] This application embodiment acquires captured video, filters candidate face images from the video, and the candidate face images are images possessing the facial features of the candidate. The candidate face images are matched with at least one target face image. If a matching face image exists among the target face images, the level information and code information of the matching person corresponding to the matching face image are obtained. Based on the level information, a target decision tree corresponding to the level information is determined from multiple decision trees. The target decision tree includes multiple decision nodes with hierarchical relationships, and each decision node corresponds to a sub-processing method. Based on the code information, a target sub-processing method is determined from the multiple sub-processing methods to obtain the processing decision for the matching person. In this way, dangerous individuals are automatically identified through face recognition, and a response decision is made for dangerous individuals through a decision tree, which can avoid human monitoring and improve the intelligence level and accuracy of monitoring. Attached Figure Description

[0047] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1a This is a schematic diagram of a system for processing target persons provided in an embodiment of this application.

[0049] Figure 1b This is a flowchart illustrating the method for processing a target person as provided in an embodiment of this application.

[0050] Figure 1c The pixel value of each pixel in the segmented image provided in the embodiments of this application.

[0051] Figure 1d The segmented image after binary processing is provided in the embodiments of this application.

[0052] Figure 1e This is a schematic diagram of a decision tree provided in an embodiment of this application.

[0053] Figure 2 This is a schematic diagram of the structure of the processing device for the target person provided in the embodiments of this application.

[0054] Figure 3 This is a schematic diagram of the server structure provided in an embodiment of this application. Detailed Implementation

[0055] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0056] This application provides a method, apparatus, and computer-readable storage medium for processing a target person.

[0057] Please see Figure 1a , Figure 1aThis is a schematic diagram of a target person processing system provided in an embodiment of this application. The system may include at least one camera device 1000, at least one server 2000, at least one database 3000, and a network 4000. The camera device 1000 can be a camera, electronic eye, or mobile terminal, etc. The camera device 1000 can connect to the server 2000, or multiple servers 2000, via the network 4000, thereby synchronizing or transmitting the captured video to the server 2000 via the network 4000. The network 4000 can be a wireless network or a wired network, such as a wireless local area network (WLAN), local area network (LAN), cellular network, 2G network, 3G network, 4G network, 5G network, etc. In addition, different camera devices 1000 can also connect to the server 2000 using their own Bluetooth network or hotspot network. Furthermore, the system may include a database 3000, which can be used to store information such as the target person's facial image for the target task.

[0058] This application provides a method for processing a target person, which can be executed by a server. For example... Figure 1a As shown, the server 2000 acquires video footage from the camera device 1000, filters candidate face images from the video footage (the candidate face images are images with the facial features of the candidate), matches the candidate face images with at least one target face image, if a matching face image exists in the target face image, acquires the level information and code information of the matching person corresponding to the matching face image, determines a target decision tree corresponding to the level information from multiple decision trees based on the level information, the target decision tree includes multiple decision nodes with hierarchical relationships, each decision node corresponds to a sub-processing method, and determines the target sub-processing method from the multiple sub-processing methods based on the code information, thus obtaining the processing decision for the matching person. Based on this, dangerous individuals can be automatically identified through face recognition, and a response decision can be made for dangerous individuals through a decision tree, avoiding manual monitoring and improving the intelligence and accuracy of monitoring.

[0059] It should be noted that, Figure 1a The schematic diagram of the target person processing system shown is merely an example. The target person processing system and scenario described in this application embodiment are for the purpose of more clearly illustrating the technical solutions of this application embodiment and do not constitute a limitation on the technical solutions provided in this application embodiment. As those skilled in the art will know, with the evolution of the target person processing system and the emergence of new business scenarios, the technical solutions provided in this application embodiment are also applicable to similar technical problems.

[0060] In this embodiment, the description will be from the perspective of the target person's processing device, which can be integrated into a computer device that has a storage unit and a microprocessor and thus computing power.

[0061] Please see Figure 1b , Figure 1b This is a flowchart illustrating a method for processing a target person as provided in an embodiment of this application. The method for processing the target person includes:

[0062] In step 101, a video is captured, and candidate face images are selected from the video. The candidate face images are images that have the facial features of the candidate.

[0063] The video can be captured by a camera device and then transmitted or synchronized to the server, thus enabling the server to obtain the captured video.

[0064] Specifically, after acquiring the captured video, candidate face images can be selected from it. These candidate face images are images that contain the facial features of the candidate individuals. The selection process can involve segmenting the captured video into frames, obtaining each frame, and then selecting images containing the facial features of the candidate individuals from each frame to obtain the candidate face images.

[0065] In some implementations, after the step of filtering candidate face images from the captured video, the method further includes:

[0066] At least two facial feature vectors are extracted from the candidate face image to obtain the candidate facial feature group of the candidate face image.

[0067] In addition, at least two facial feature vectors can be extracted from the candidate face image to obtain the candidate face feature group of the candidate face image.

[0068] Specifically, facial features can be feature vectors for facial key points, facial expressions, and facial age. Facial key points include features such as eyes, ears, mouth, and nose; facial expressions include expressions such as crying and smiling; facial age includes features such as 18 years old and 22 years old. The above facial features are just examples, and the types of facial features are not limited here.

[0069] In some implementations, the step of extracting at least two facial feature vectors from the candidate face image includes:

[0070] (1.1) Collect at least two facial features from the candidate face image;

[0071] (1.2) Perform image segmentation on each facial feature to obtain multiple segmented images;

[0072] (1.3) Perform binary processing on the pixel values ​​of each pixel in the segmented image to obtain the segmented image after binary processing;

[0073] (1.4) Calculate the feature vector of the processed segmented image to obtain at least two facial feature vectors.

[0074] One method for extracting facial feature vectors from candidate face images is using Local Binary Pattern (LBP). Specifically, an LBP operator is defined, typically within a 3×3 window, to segment facial features and obtain a segmented image. For example, if the facial feature is the eyes, a 3×3 window is used to segment the eye area from the candidate face image.

[0075] Specifically, after obtaining the segmented image, the pixel values ​​of the pixels within the segmented image need to be binary processed to obtain a binary segmented image consisting of 0s or 1s. The binary segmented image is a matrix sequence, and the feature vector of the processed segmented image is calculated by drawing lines from the middle pixel as endpoints in a direction denoted as '1', thus forming the feature vector.

[0076] In some implementations, the step of performing binary processing on the pixel values ​​of each segmented image to obtain a binary-processed segmented image includes:

[0077] (1.1) Obtain the pixel value of the center pixel of the segmented image, and calculate the pixel difference between each other pixel in the segmented image and the center pixel, wherein the other pixels are the pixels in the segmented image other than the center pixel;

[0078] (1.2) Pixels whose pixel difference is within the preset pixel difference range are defined as 1, and pixels whose pixel difference is not within the preset pixel difference range are defined as 0, to obtain the segmented image after binary processing.

[0079] The binary processing method involves calculating the pixel difference between other pixels in the segmented image and the center pixel, and setting pixels within a preset pixel difference range as 1 and pixels outside the preset pixel difference range as 0, thus obtaining the segmented image after binary processing.

[0080] For example, such as Figure 1c As shown, Figure 1cThis refers to the pixel value of each pixel in the segmented image provided in this embodiment. Since a 3×3 LBP operator is used to segment the image, the segmented image is also composed of 3×3 pixels. The pixel value of the center pixel is 83. The pixel difference between each other pixel and the center pixel is calculated sequentially. Pixels with pixel differences within the range [0, 256] are recorded as 1, and pixels not within this range are recorded as 0, resulting in... Figure 1d , Figure 1d The segmented image after binary processing is provided in the embodiments of this application.

[0081] In step 102, the candidate face image is matched with at least one target face image.

[0082] The system can pre-store target facial images of the target person, which are then used to match candidate facial images with at least one target facial image to determine whether a candidate facial image matches at least one target facial image. If a match is found, the candidate is confirmed to be the target person; otherwise, the candidate is confirmed to be the target person. The target person refers to specific individuals, such as dangerous individuals, but this is not a specific limitation.

[0083] In some implementations, the step of matching the candidate face image with at least one target face image includes:

[0084] (1) Match the image similarity between the candidate face image and at least one target face image in the target face image group;

[0085] (2) Target face images with image similarity greater than a preset similarity threshold are identified as candidate matching face images;

[0086] (3) Obtain the candidate matching face feature group of the candidate matching face image. If the candidate face feature group is the same as the candidate matching face feature group, then the candidate matching face image is determined as the matching face image.

[0087] The method for matching candidate face images with at least one target face image can be as follows: match the image similarity between the candidate face image and at least one target face image in the target face image group; and determine the target face image with an image similarity greater than a preset similarity threshold as a candidate matching face image. To improve the accuracy of matching, the candidate matching face feature group and the candidate face feature group of the candidate matching face image can also be compared. If they are the same, the candidate matching face image is determined as the matching face image. Thus, whether the candidate is the target person is determined based on both image similarity and comparison of face feature groups, thereby improving the recognition accuracy. Specifically, the preset similarity threshold can be 95%, etc., and is not limited here.

[0088] In some implementations, prior to the step of matching the candidate face image with at least one target face image, the method further includes:

[0089] (1) The candidate face image is preprocessed to obtain the preprocessed candidate face image;

[0090] (2) The preprocessed candidate face image is corrected using a preset correction model to obtain the corrected candidate face image.

[0091] Since the candidate face images are captured by camera equipment, and camera lenses are generally wide-angle lenses, the faces captured will be distorted compared to those captured by a flat lens. Therefore, before matching, the candidate face images can be preprocessed to obtain preprocessed candidate face images, and then corrected using a preset correction model to obtain corrected candidate face images.

[0092] Specifically, the correction model corrects the preprocessed candidate face image by performing a perspective transformation. The equations of the perspective transformation have 8 unknowns, and 4 sets of mapping points are found. The four points determine a three-dimensional space.

[0093] In some embodiments, the step of preprocessing the candidate face image to obtain a preprocessed candidate face image includes:

[0094] The candidate face image is subjected to grayscale processing and edge detection processing to obtain a preprocessed candidate face image.

[0095] The preprocessing steps include grayscale processing and edge detection processing. The grayscale processing is used to convert the candidate face image to grayscale, and the edge detection processing is used to detect the edges of the face from the candidate face image.

[0096] In step 103, if there is a matching face image in the target face image that matches the candidate face image, the level information and code information of the matching person corresponding to the matching face image are obtained.

[0097] When the target face image is entered into the database, the target face image corresponding to each target person will be marked with a level and a code.

[0098] Specifically, taking a target person as an example, when the target person's facial image is entered into the database, the facial image corresponding to each dangerous person will be marked with a danger level, such as Level 1 danger, Level 2 danger, etc., and the facial image corresponding to the dangerous person will be marked with a code, such as 07214975.

[0099] In step 104, a target decision tree corresponding to the level information is determined from multiple decision trees based on the level information. The target decision tree includes multiple decision nodes with hierarchical relationships, and each decision node corresponds to a sub-processing method.

[0100] Among them, such as Figure 1e As shown, Figure 1e This is a schematic diagram of a decision tree provided in an embodiment of this application. Each decision tree includes multiple decision nodes with a hierarchical relationship, for example... Figure 1e The decision tree consists of decision nodes 11 at the first level, 21 and 22 at the second level, and 31, 32, 33, 34, and 35 at the third level. Different decision trees can be set up for different levels of information. This is because the decision-making process of the decision tree proceeds from the highest level to the next, for example, from decision node 11 to 21 and finally to 32. For higher-level targets, the corresponding sub-processing method requires multiple major decisions to be on the same branch of the decision tree, such as decision nodes 11, 21, and 32 on the same branch. For lower-level targets, the corresponding sub-processing method can set multiple major decisions on different branches of the decision tree, such as decision nodes 21 and 22 on different branches.

[0101] For example, for a Level 1 (highest level) target, decision node 11 corresponds to a warning, decision node 21 corresponds to a screen notification, and decision node 32 corresponds to an alarm. For a Level 3 (lowest level) target, decision node 11 corresponds to a warning, decision node 21 corresponds to a screen notification, and decision node 22 corresponds to an alarm. By setting the alarm sub-processing method on decision node 22, which is on a different branch from decision node 21, the simultaneous occurrence of both screen notifications and alarms is avoided for Level 3 targets. However, for Level 1 targets, both screen notifications and alarms can occur simultaneously.

[0102] In step 105, the target sub-processing method is determined from the plurality of sub-processing methods based on the code information, and the processing decision for the matched person is obtained.

[0103] Specifically, decision-making is carried out in the target decision tree through code information, thereby determining the target sub-processing method from the multiple sub-processing methods, and obtaining the processing decision of the matching person.

[0104] In some implementations, the code information includes at least one sub-code information segment, each sub-code information segment corresponding to a decision node level, and each decision node in the decision node level is assigned a code range. The step of determining the target sub-processing method from the plurality of sub-processing methods based on the code information to obtain the processing decision for the matched person includes:

[0105] (1) Determine the target sub-processing method of the matched person at each decision node level based on the code range corresponding to the decision node in each decision node level and each segment of sub-code information in the code information.

[0106] (2) Based on the target sub-processing method, the processing method of the matched person is obtained.

[0107] The code information includes at least one sub-code segment. For example, code information 07214975 consists of four sub-code segments: 07, 21, 49, and 75. Each sub-code segment corresponds to a decision node level. For instance, sub-code 07 corresponds to the first-level decision node level, sub-code 21 to the second-level, sub-code 49 to the third-level, and sub-code 75 to the fourth-level. Each decision node within a decision node level has a corresponding code range, for example... Figure 1e The code range for decision node 21, which is at the secondary decision node level, is 31–50, and the code range for decision node 22 is 1–30. Therefore, based on the code range corresponding to each decision node at each decision node level, and each segment of sub-code information in the code information, the target sub-processing method for the matched person at each decision node level is determined. And based on the target sub-processing method, the processing method for the matched person is obtained.

[0108] For example, the target sub-processing method at the first-level decision node level is a warning, the target sub-processing method at the second-level decision node level is a screen prompt, and the target sub-processing method at the third-level decision node level is an alarm. The processing method for the matched person is a warning + screen prompt + alarm.

[0109] In some implementations, the step of determining the target sub-processing method of the matched person at each decision node level based on the code range corresponding to the decision node in each decision node level and each segment of sub-code information in the code information includes:

[0110] (1.1) Obtain the target sub-codes included in the sub-code information in the code information, and the code range of each decision node in the decision node hierarchy corresponding to the sub-code information;

[0111] (1.2) The decision node corresponding to the code range including the target sub-code is determined as the target decision node, and the sub-processing method corresponding to the target decision node is determined as the target sub-processing method, so as to obtain the target sub-processing method of the matched person at each decision node level.

[0112] The method for determining the target sub-processing method for the matched person at each decision node level is as follows: Obtain the target sub-code from the sub-code information, for example, the target sub-code in the sub-code information is "21". Then, obtain the code range of each decision node in the decision node level corresponding to the sub-code information. For example, the sub-code information 21 corresponds to the second-level decision node level, and the code range of decision node 21 in the second-level decision node level is 31-50, while the code range of decision node 22 is 1-30. The decision node corresponding to the code range including the target sub-code is determined as the target decision node, and the sub-processing method corresponding to the target decision node is determined as the target sub-processing method. For example, the target sub-code "21" is within the code range of decision node 22 (31-50), so the target node for the matched person at the second-level decision node level is decision node 22, and the sub-processing method corresponding to decision node 22 is determined as the target sub-processing method. This process is repeated to obtain the target sub-processing method for the matched person at each decision node level.

[0113] Specifically, the code range can be entropy. In the decision tree, all decision nodes in the same branch except for the lowest-level decision node are assigned entropy. Entropy can be used to train the decision tree to improve decision accuracy.

[0114] Since the code information is manually marked, and different target individuals must have different code information during the marking process, the target individual's code information can be deduced from their processing decisions and the code range of each decision node. For example, the processing decision for target individual A should be: warning + screen prompt + alarm, and... Figure 1e In the first-level decision node hierarchy, decision node 11 corresponds to a warning; in the second-level decision node hierarchy, decision node 22 corresponds to a screen notification; and in the third-level decision node hierarchy, decision node 34 corresponds to an alarm. The code range for decision node 11 is 10-50, for decision node 22 it is 1-20, and for decision node 34 it is 60-80. Therefore, based on the warning + screen notification + alarm, we can deduce that the code information for target person A consists of any code from 10-50, any code from 1-20, and any code from 60-80, for example, 361979.

[0115] Specifically, the range of codes for each decision node in the decision tree can be optimized by training the decision tree. This is done by selecting a certain number of codes and their corresponding processing decisions as a training set to train the decision tree, and selecting a certain number of codes and their corresponding processing decisions as a validation set to validate the trained decision tree. Once the validation is successful, the range of codes for each decision node is determined.

[0116] The advantage of using a decision tree compared to setting a mapping relationship between each code and the processing decision is that if the number of target persons is large, it is necessary to search for the corresponding processing method for the matching face image from a large number of mapping relationships when determining the processing method for the matching face image, which takes a long time. However, by using a decision tree, the processing method for the original large number of target persons is simplified through a multi-level and multi-node approach. The processing method for the matching face image is determined by a decision tree composed of a smaller number of levels and nodes based on the code information, which is simpler and takes less time.

[0117] As described above, this embodiment of the application acquires a captured video, filters candidate face images from the video, and the candidate face images are images possessing the facial features of the candidate. The candidate face images are then matched with at least one target face image. If a matching face image exists among the target face images, the level information and code information of the matching person corresponding to the matching face image are obtained. Based on the level information, a target decision tree corresponding to the level information is determined from multiple decision trees. The target decision tree includes multiple decision nodes with hierarchical relationships, and each decision node corresponds to a sub-processing method. Based on the code information, a target sub-processing method is determined from the multiple sub-processing methods to obtain the processing decision for the matching person. In this way, dangerous individuals are automatically identified through face recognition, and a response decision is made for these dangerous individuals through a decision tree, avoiding manual monitoring and improving the intelligence and accuracy of monitoring.

[0118] To facilitate better implementation of the target person processing method provided in the embodiments of this application, this application also provides an apparatus based on the aforementioned target person processing method. The meanings of the terms used are the same as in the aforementioned target person processing method, and specific implementation details can be found in the descriptions within the method embodiments.

[0119] Please see Figure 2 , Figure 2 This is a schematic diagram of the structure of a target person processing device provided in an embodiment of this application. The target person processing device may include a filtering module 301, a matching module 302, an acquisition module, a first determination module 304, and a second determination module 305, etc.

[0120] The filtering module 301 is used to acquire the captured video and filter out candidate face images from the captured video, wherein the candidate face images are images that have the facial features of the candidate.

[0121] Matching module 302 is used to match the candidate face image with at least one target face image;

[0122] The acquisition module 303 is used to acquire the level information and code information of the matching person corresponding to the matching face image if there is a matching face image in the target face image that matches the candidate face image;

[0123] The first determining module 304 is used to determine the target decision tree corresponding to the level information from multiple decision trees according to the level information. The target decision tree includes multiple decision nodes with hierarchical relationship, and each decision node corresponds to a sub-processing method.

[0124] The second determining module 305 is used to determine the target sub-processing method from the plurality of sub-processing methods based on the code information, and to obtain the processing decision of the matching person.

[0125] In some embodiments, the code information includes at least one sub-code information segment, each sub-code information segment corresponding to a decision node level, and each decision node in the decision node level is assigned a code range. The second determining module 305 includes:

[0126] The first determining submodule is used to determine the target subprocessing method of the matched person at each decision node level based on the code range corresponding to the decision node in each decision node level and each segment of sub-code information in the code information.

[0127] The second determining submodule is used to determine the processing method of the matched person based on the target subprocessing method.

[0128] In some implementations, the first determining module 304 includes:

[0129] The acquisition unit is used to acquire the target sub-codes included in the sub-code information in the code information, and the code range of each decision node in the decision node hierarchy corresponding to the sub-code information.

[0130] The first determining unit is used to determine the decision node corresponding to the code range including the target sub-code as the target decision node, and to determine the sub-processing method corresponding to the target decision node as the target sub-processing method, so as to obtain the target sub-processing method of the matched person at each decision node level.

[0131] In some embodiments, the apparatus further includes:

[0132] An extraction module is used to extract at least two types of facial feature vectors from the candidate face image to obtain a candidate facial feature group of the candidate face image;

[0133] The matching module 302 includes:

[0134] The matching submodule is used to match the image similarity between the candidate face image and at least one target face image in the target face image group;

[0135] The third determination submodule is used to determine target face images with image similarity greater than a preset similarity threshold as candidate matching face images;

[0136] The fourth determination submodule is used to obtain the candidate matching facial feature group of the candidate matching face image. If the candidate facial feature group is the same as the candidate matching facial feature group, then the candidate matching face image is determined as the matching face image.

[0137] In some embodiments, the extraction module includes:

[0138] The acquisition submodule is used to acquire at least two facial features from the candidate face image;

[0139] The segmentation submodule is used to segment the image for each facial feature, resulting in multiple segmented images;

[0140] The binary processing submodule is used to perform binary processing on the pixel values ​​of each pixel in the segmented image to obtain the segmented image after binary processing.

[0141] The calculation submodule is used to calculate the feature vector of the processed segmented image to obtain at least two kinds of facial feature vectors.

[0142] In some implementations, the binary processing submodule includes:

[0143] The calculation unit is used to obtain the pixel value of the center pixel of the segmented image and calculate the pixel difference between each other pixel in the segmented image and the center pixel.

[0144] The second determining unit is used to determine pixels whose pixel differences are within a preset pixel difference range as 1, and pixels whose pixel differences are not within the preset pixel difference range as 0, to obtain a segmented image after binary processing.

[0145] In some embodiments, the apparatus further includes:

[0146] The preprocessing module is used to preprocess the candidate face image to obtain a preprocessed candidate face image;

[0147] The correction module is used to correct the preprocessed candidate face image using a preset correction model to obtain the corrected candidate face image.

[0148] In some embodiments, the preprocessing module includes:

[0149] The grayscale processing and edge detection processing submodule is used to perform grayscale processing and edge detection processing on the candidate face image to obtain a preprocessed candidate face image.

[0150] As described above, this embodiment of the application acquires a captured video through a filtering module 301, filters out candidate face images from the video, and the candidate face images are images possessing the facial features of the candidate; a matching module 302 matches the candidate face images with at least one target face image; an acquisition module 303, if a matching face image exists in the target face image that matches the candidate face image, acquires the level information and code information of the matching person corresponding to the matching face image; a first determination module 304 determines a target decision tree corresponding to the level information from multiple decision trees based on the level information, the target decision tree including multiple decision nodes with hierarchical relationships, each decision node corresponding to a sub-processing method; a second determination module 305 determines a target sub-processing method from the multiple sub-processing methods based on the code information, and obtains the processing decision for the matching person. In this way, dangerous individuals are automatically identified through face recognition, and response decisions are made for dangerous individuals through decision trees, avoiding manual monitoring and improving the intelligence level and regulatory accuracy of monitoring.

[0151] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0152] Accordingly, embodiments of this application also provide a server, such as Figure 3 As shown, Figure 3 This is a schematic diagram of the structure of a server provided in an embodiment of this application. The server 2000 includes a processor 401 with one or more processing cores, a memory 402 with one or more computer-readable storage media, and a computer program stored on the memory 402 and executable on the processor. The processor 401 and the memory 402 are electrically connected. Those skilled in the art will understand that the server structure shown in the figure does not constitute a limitation on the server, and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0153] The processor 401 is the control center of the server 2000. It connects various parts of the server 2000 through various interfaces and lines. By running or loading software programs and / or modules stored in the memory 402, and calling data stored in the memory 402, it performs various functions of the server 2000 and processes data, thereby monitoring the server 2000 as a whole.

[0154] In this embodiment, the processor 401 in the server 2000 loads the instructions corresponding to the processes of one or more applications into the memory 402 according to the following steps, and the processor 401 runs the applications stored in the memory 402 to achieve various functions:

[0155] The process involves acquiring a video recording, filtering candidate face images from the video (each candidate face image is an image containing the facial features of a candidate), matching the candidate face images with at least one target face image, and if a matching face image exists among the target face images, acquiring the level information and code information of the matching person corresponding to the matching face image. Based on the level information, a target decision tree corresponding to the level information is determined from multiple decision trees. The target decision tree includes multiple decision nodes with hierarchical relationships, each decision node corresponding to a sub-processing method. Based on the code information, a target sub-processing method is determined from the multiple sub-processing methods to obtain the processing decision for the matching person.

[0156] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0157] Optional, such as Figure 3 As shown, the server 2000 also includes an input unit 403 and a power supply 404. The processor 401 is electrically connected to both the input unit 403 and the power supply 404. Those skilled in the art will understand that... Figure 3 The server structure shown does not constitute a limitation on the server and may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0158] The input unit 403 can be used to receive input numbers, characters, or user characteristic information (such as fingerprints, iris, facial information, etc.), and to generate keyboard, mouse, joystick, optical, or trackball signal inputs related to user settings and function control.

[0159] Power supply 404 is used to power the various components of server 2000. Optionally, power supply 404 can be logically connected to processor 401 through a power management system, thereby enabling functions such as charging, discharging, and power consumption management through the power management system. Power supply 404 may also include one or more DC or AC power supplies, recharging systems, power fault detection circuits, power converters or inverters, power status indicators, and other arbitrary components.

[0160] although Figure 3 As not shown in the diagram, the server 2000 may also include a camera, sensors, a wireless fidelity module, a Bluetooth module, etc., which will not be described in detail here.

[0161] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.

[0162] As described above, the server provided in this embodiment can acquire captured videos, filter candidate face images from the videos (the candidate face images are images with the facial features of the candidate), match the candidate face images with at least one target face image, and if a matching face image exists in the target face image that matches the candidate face image, obtain the level information and code information of the matching person corresponding to the matching face image; determine the target decision tree corresponding to the level information from multiple decision trees based on the level information, the target decision tree including multiple decision nodes with hierarchical relationships, each decision node corresponding to a sub-processing method; determine the target sub-processing method from the multiple sub-processing methods based on the code information, and obtain the processing decision for the matching person. Thus, dangerous individuals are automatically identified through face recognition, and response decisions are made for dangerous individuals through decision trees, avoiding manual monitoring and improving the intelligence and accuracy of monitoring.

[0163] Those skilled in the art will understand that all or part of the steps in the various methods of the above embodiments can be performed by instructions, or by instructions controlling related hardware. These instructions can be stored in a computer-readable storage medium and loaded and executed by a processor.

[0164] Therefore, embodiments of this application provide a computer-readable storage medium storing a plurality of computer programs that can be loaded by a processor to execute steps in any of the target person processing methods provided in embodiments of this application. For example, the computer program can execute the following steps:

[0165] The process involves acquiring a video recording, filtering candidate face images from the video (each candidate face image is an image containing the facial features of a candidate), matching the candidate face images with at least one target face image, and if a matching face image exists among the target face images, acquiring the level information and code information of the matching person corresponding to the matching face image. Based on the level information, a target decision tree corresponding to the level information is determined from multiple decision trees. The target decision tree includes multiple decision nodes with hierarchical relationships, each decision node corresponding to a sub-processing method. Based on the code information, a target sub-processing method is determined from the multiple sub-processing methods to obtain the processing decision for the matching person.

[0166] For details on the implementation of each of the above operations, please refer to the previous examples, which will not be repeated here.

[0167] The storage medium may include: read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0168] Since the computer program stored in the storage medium can execute the steps in any of the target person processing methods provided in the embodiments of this application, the beneficial effects that any of the target person processing methods provided in the embodiments of this application can achieve can be realized, as detailed in the preceding embodiments, and will not be repeated here.

[0169] The foregoing has provided a detailed description of a method, apparatus, computer-readable storage medium, and server for processing a target person, as provided in the embodiments of this application. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, there will be changes in the specific implementation methods and application scope based on the ideas of this application. Therefore, the content of this specification should not be construed as a limitation of this application.

Claims

1. A processing method of a target person, characterized by, The method comprises the following steps: acquiring a shooting video, and screening a candidate face image from the shooting video, the candidate face image being an image with face features of a candidate character; matching the candidate face image with at least one target face image; if there is a matching face image matching the candidate face image in the target face image, acquiring grade information and code information of a matching character corresponding to the matching face image, wherein the code information comprises at least one piece of sub-code information, each piece of sub-code information corresponding to a decision node level, and each decision node in each decision node level is provided with a code range; determining a target decision tree corresponding to the grade information from a plurality of decision trees according to the grade information, the target decision tree comprising a plurality of decision nodes with a hierarchical relationship, and each decision node corresponds to a sub-processing mode; determining a target sub-processing mode of the matching character in each decision node level according to the code range of the decision node in each decision node level and each piece of sub-code information in the code information; obtaining a processing mode of the matching character based on the target sub-processing mode.

2. The method of processing a target person according to claim 1, wherein The step of determining the target sub-processing mode of the matching character in each decision node level according to the code range of the decision node in each decision node level and each piece of sub-code information in the code information comprises: acquiring a target sub-code included in the sub-code information in the code information and a code range of each decision node in the decision node level corresponding to the sub-code information; determining a decision node corresponding to the code range of the target sub-code as a target decision node, and determining a sub-processing mode corresponding to the target decision node as a target sub-processing mode to obtain the target sub-processing mode of the matching character in each decision node level.

3. The method of processing a target person according to claim 1, wherein After the step of screening the candidate face image from the shooting video, the method further comprises: extracting at least two face feature vectors from the candidate face image to obtain a candidate face feature group of the candidate face image; The step of matching the candidate face image with at least one target face image comprises: matching the candidate face image with at least one target face image in a target face image group in terms of image similarity; determining a target face image with an image similarity greater than a preset similarity threshold as a candidate matching face image; acquiring a candidate matching face feature group of the candidate matching face image, and determining the candidate matching face image as a matching face image if the candidate face feature group is identical to the candidate matching face feature group.

4. The method of processing a target person according to claim 3, wherein The step of extracting at least two face feature vectors from the candidate face image comprises: collecting at least two face features in the candidate face image; performing image segmentation on each face feature to obtain a plurality of segmented images; performing binary processing on pixel values of pixels in each segmented image to obtain a binary-processed segmented image; calculating a feature vector of the processed segmented image to obtain at least two face feature vectors.

5. The method of processing a target person according to claim 4, wherein The step of performing binary processing on pixel values of pixels in each segmented image to obtain a binary-processed segmented image comprises: obtaining a pixel value of a center pixel point of the segmented image, and calculating a pixel difference value between each other pixel point and the center pixel point in the segmented image, the other pixel point being a pixel point other than the center pixel point in the segmented image; determining a pixel within a preset pixel difference value range as 1, and determining a pixel not within the preset pixel difference value range as 0, to obtain a binary-processed segmented image.

6. The method of processing a target person according to Claim 1, wherein Before the step of matching the candidate face image with at least one target face image, the method further comprises: preprocessing the candidate face image to obtain a preprocessed candidate face image; correcting the preprocessed candidate face image using a preset correction model to obtain a corrected candidate face image.

7. The method of processing a target person according to claim 6, wherein The step of preprocessing the candidate face image to obtain a preprocessed candidate face image comprises: performing grayscale processing and edge detection processing on the candidate face image to obtain a preprocessed candidate face image.

8. A processing device of a target person, characterized by, The method comprises: a screening module configured to obtain a shooting video, and screen a candidate face image from the shooting video, the candidate face image being an image having face features of a candidate character; a matching module configured to match the candidate face image with at least one target face image; an obtaining module configured to, if there is a matching face image matching the candidate face image in the target face image, obtain level information and code information of a matching character corresponding to the matching face image, wherein the code information comprises at least one piece of sub-code information, each piece of sub-code information corresponding to a decision node level, and each decision node in each decision node level is provided with a code range; a first determining module configured to determine a target decision tree corresponding to the level information from a plurality of decision trees according to the level information, the target decision tree comprising a plurality of decision nodes having a hierarchical relationship, and each decision node corresponds to a sub-processing mode; a second determining module configured to determine a target sub-processing mode of the matching character in each decision node level according to a code range corresponding to a decision node in each decision node level and each piece of sub-code information in the code information, and obtain a processing mode of the matching character based on the target sub-processing mode.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a plurality of instructions adapted to be loaded by the processor to execute the steps in the processing method of the target character according to any one of claims 1 to 7.

10. A server comprising a memory, a processor and a computer program stored on the memory and running on the processor, characterized in that, The processor executes the program to implement the steps in the processing method of the target character according to any one of claims 1 to 7.

Images