WO2020124710A1 - 安检辅助分析方法、装置、计算机设备和存储介质 - Google Patents

安检辅助分析方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2020124710A1
WO2020124710A1 PCT/CN2019/071257 CN2019071257W WO2020124710A1 WO 2020124710 A1 WO2020124710 A1 WO 2020124710A1 CN 2019071257 W CN2019071257 W CN 2019071257W WO 2020124710 A1 WO2020124710 A1 WO 2020124710A1
Authority
WO
WIPO (PCT)
Prior art keywords
image
feature
micro
emotion
identity
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.)
Ceased
Application number
PCT/CN2019/071257
Other languages
English (en)
French (fr)
Inventor
万梅
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
OneConnect Smart Technology Co Ltd
Original Assignee
OneConnect Smart Technology Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by OneConnect Smart Technology Co Ltd filed Critical OneConnect Smart Technology Co Ltd
Publication of WO2020124710A1 publication Critical patent/WO2020124710A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V10/00—Arrangements for image or video recognition or understanding
    • G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
    • G06V10/74—Image or video pattern matching; Proximity measures in feature spaces
    • G06V10/75—Organisation of the matching processes, e.g. simultaneous or sequential comparisons of image or video features; Coarse-fine approaches, e.g. multi-scale approaches; using context analysis; Selection of dictionaries
    • G—PHYSICS
    • G07—CHECKING-DEVICES
    • G07C—TIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
    • G07C9/00—Individual registration on entry or exit

Definitions

  • the present application relates to a security analysis auxiliary analysis method, device, computer equipment and storage medium.
  • the existing security inspection equipment mainly checks luggage items carried by people or carries out a full body inspection on the human body to check whether the human body carries contraband, but these cannot actively identify criminals through the characteristics of criminals.
  • the security inspector can only see the X-ray image of the item, the current counter number and other information on the display screen, and cannot fully understand the situation of the security inspection site. Only the one-sided information can be used for security inspection. There may be a missing inspection or Misdetection and other phenomena lead to low accuracy of security inspection, poor security inspection effect, and passive prevention of danger.
  • a security inspection auxiliary analysis method, apparatus, computer equipment, and storage medium are provided.
  • a security analysis auxiliary analysis method including:
  • image features are extracted from the real-time security image, and the image features are Similarity comparison of detailed features of facial expressions to obtain micro-expression similarity;
  • the emotion pattern of the security inspection object is determined according to the emotion value.
  • a security analysis auxiliary analysis device including:
  • a security inspection acquisition module used to obtain the identification information and real-time security inspection images of the security inspection object
  • An identity image search module used to search for an identity image corresponding to the identity identification information
  • a portrait comparison module configured to perform portrait comparison between the real-time security inspection image and the identity image to obtain a portrait comparison result
  • the micro-expression analysis module is used to extract image features from the real-time security image when it is determined that the portrait corresponding to the real-time security image and the identity corresponding to the identity image belong to the same person according to the portrait comparison result. Get the similarity of the micro-expressions by describing the image features and the detailed features of the preset micro-expressions;
  • the expression coefficient acquisition module is used to search for the expression coefficient corresponding to each of the preset micro expressions in each emotion mode
  • a calculation module configured to calculate the emotion value of the security inspection object in each emotion mode according to the expression coefficient and the micro-expression similarity
  • the emotion pattern determination module is used to determine the emotion pattern of the security inspection object according to the emotion value.
  • a computer device includes a memory and one or more processors.
  • the memory stores computer-readable instructions.
  • the steps of the security inspection auxiliary analysis method provided in any embodiment of the present application are implemented.
  • One or more non-volatile computer-readable storage media storing computer-readable instructions, which when executed by one or more processors, cause the one or more processors to implement any one of the embodiments of the present application.
  • FIG. 1 is an application scenario diagram of a security inspection auxiliary analysis method according to one or more embodiments.
  • FIG. 2 is a schematic flowchart of a security analysis auxiliary analysis method according to one or more embodiments.
  • FIG. 3 is a schematic flowchart of a step of comparing portraits according to one or more embodiments.
  • FIG. 4 is a schematic flowchart of a step of acquiring security inspection information according to one or more embodiments.
  • FIG. 5 is a schematic flowchart of a micro-expression classification step according to one or more embodiments.
  • FIG. 6 is a block diagram of a security inspection auxiliary analysis device according to one or more embodiments.
  • FIG. 7 is a block diagram of a computer device in accordance with one or more embodiments.
  • the security analysis auxiliary analysis method provided in this application can be applied to the application environment shown in FIG. 1.
  • the terminal 102 communicates with the server 104 via the network.
  • the terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablets, and portable security inspection smart devices.
  • the server 104 may be implemented by an independent server or a server cluster composed of multiple servers.
  • the terminal 102 is a security inspection auxiliary analysis device installed at a location with a large passenger flow such as a train station or an airport.
  • the security inspection auxiliary analysis device can not only obtain the real-time security inspection image of the security inspection object indirectly by obtaining an externally installed camera, but also can use the camera module to take real-time shooting of the security inspection object to obtain the real-time security inspection image directly.
  • the terminal 102 uploads the acquired identity identification information to the server 104, and the server 104 searches for the identity image according to the received identity identification information and sends the corresponding identity image to the terminal 102. After receiving the identity image, the terminal 102 first performs real-time security inspection The image is compared with the identity image to obtain the portrait comparison result; and then the micro expression analysis is performed.
  • the server 104 may be implemented by an independent server or a server cluster composed of multiple servers to provide data access services for the terminal 102.
  • a security inspection auxiliary analysis method is provided. Taking the method applied to the terminal 102 in FIG. 1 as an example for illustration, it includes the following steps:
  • Step 202 Acquire the identification information and real-time security inspection image of the security inspection object.
  • the security inspection object may be a passenger or a cabin crew.
  • the identification information may be the ID number, passport number, fingerprint, etc. of the security inspection object stored in the security inspection system in advance.
  • the terminal 102 may obtain the ID number of the security inspection object ID card and/or obtain the fingerprint of the security inspection object.
  • the terminal 102 acquires the real-time security inspection image while acquiring the identification information of the security inspection object.
  • the real-time security inspection image is a captured image when the security inspection object passes through a predetermined position. One or more images can be taken as needed. When the captured image is only one, the captured image is the real-time security image; when there are multiple captured images, one of the captured images can be set as the real-time security image, or a certain number of captured images can be set Real-time security image.
  • Step 204 Search for the identity image corresponding to the identification information.
  • the terminal 102 sends the acquired identity identification information to the server, and searches in the server according to the identity identification information, and acquires the corresponding identity image.
  • the terminal 102 may generate an image acquisition request containing identity identification information and send it to the server, and receive the server to return the requested identity image based on the image acquisition request.
  • the server searches according to the identification information in the image acquisition request, and sends the found identification image to the terminal 102.
  • Step 206 Perform portrait comparison between the real-time security inspection image and the identity image to obtain a portrait comparison result.
  • a face image comparison recognition method can also be used, or a two-shot comparison recognition method can be used to compare the real-time security image and the identity image for portrait comparison judgment.
  • the terminal 102 first normalizes the real-time security image and the identity image; then obtains the pre-established general three-dimensional face model library, through pre-trained multi-tasking Learning neural networks to perform face detection and key point information extraction of real-time security images and identity images to determine the position of each organ, and then extract corresponding size data for face avatars in real-time security images and identity images; calculate based on size data Face ratio data in portraits; determine whether the face ratio data of identity avatars and security avatars are consistent and generate portrait comparison results.
  • the universal three-dimensional face model library is suitable for all groups of people. It stores nine areas and data for classifying faces into eyebrows, eyes, noses, mouths, chins, cheeks, foreheads, ears, and hindbrains. Each area classification defines area outline points and feature points. The classification data also records the geometric relationship between the area outline points and the feature points. Therefore, the general three-dimensional face model library can quickly locate the positions of the eyebrows, eyes, nose, mouth, chin, cheeks, forehead, and ears on the face, and then extract the size data of each organ on the face.
  • the terminal 102 separately calculates the facial proportion data of the portraits in the real-time security image and the identity image according to the size data, and compares and judges the facial proportion data of the two images.
  • the portrait corresponding to the real-time security image and the portrait corresponding to the identity image are generated; when the face ratio data is inconsistent, the portrait corresponding to the real-time security image and the portrait corresponding to the identity image are generated The result of comparing portraits of different people.
  • Step 208 when it is determined that the portrait corresponding to the real-time security image and the portrait corresponding to the identity image belong to the same person according to the portrait comparison result, image features are extracted from the real-time security image, and micro-expressions are obtained according to the image features and the detailed features of the preset micro-expressions Similarity.
  • the terminal 102 extracts image features from the real-time security image and compares the image features with the pre-stored details of the preset micro-expression details The similarity is compared to obtain the micro-expression similarity.
  • the image features extracted from the real-time security inspection image are compared with the previously stored image features of the micro-expressions "squinting eyes", “rising the corners of the mouth”, “pulling the corners of the mouth”, “blinking eyes” and “toothing” one by one. Yes, the comparison values are 40%, 80%, 0%, 10%, and 90%, respectively. Therefore, there are only micro-expressions "squinting eyes”, “rising mouth corners”, “blinking eyes” and “toothed teeth” in real-time security inspection images, and the micro-expression similarities are 40%, 80%, 10% and 90%, respectively.
  • Step 210 Find the expression coefficient corresponding to each micro-expression under each emotion mode.
  • the terminal searches for corresponding expression coefficients in each emotion mode according to the micro expressions.
  • there are seven emotion patterns namely happy, sad, scared, angry, disgusted, surprised, and contempt.
  • the correlation between different micro-expressions and different emotion patterns is different, so the expression coefficients are also different.
  • the expression coefficients of the micro-expression "rising the corner of the mouth" correspond to happiness, sadness, fear, anger, disgust, surprise, and contempt. It is 9, 0, 0, 5, 3, 1, and 2.
  • Step 212 Calculate the emotion value of the security inspection object in each emotion mode according to the expression coefficient and the micro-expression similarity.
  • Step 214 Determine the emotion pattern of the security inspection object according to the emotion value.
  • the terminal sets the emotion mode with the largest emotion value as the emotion mode of the security inspection object according to the emotion value of each emotion mode.
  • the identity image of the security inspection object and the real-time security inspection image are obtained first, and the two images are compared and analyzed to determine the identity of the security inspection object, so that the whereabouts of the security inspection object can be traced, and fraud can also be reduced.
  • the probability of occurrence of the identity of others is obtained.
  • the facial expression of the security inspection object is micro-expressed to obtain the current emotion pattern of the security inspection object, thereby determining the real-time emotion of the security inspection object, and further realizing the understanding of the security inspection object, and assisting in security inspection decision-making. Improve the accuracy of security inspection and security inspection effect.
  • the real-time security inspection image is compared with the identity image to obtain a portrait comparison result, which specifically includes the following steps:
  • Step 302 Extract security inspection portrait information from the real-time security inspection image, and extract identity portrait information from the identity image.
  • the terminal performs portrait detection on the real-time security inspection image, determines the position of the portrait in the real-time security inspection image, and then extracts security inspection portrait information from the real-time security inspection image according to the determined image location.
  • the terminal also performs portrait detection on the identity image to determine the position of the portrait in the identity image, and then extracts identity portrait information from the identity image according to the determined image position.
  • Step 304 Extract the identity feature vector from the identity portrait information, and extract the security feature vector from the security portrait information.
  • both the identity feature vector and the security feature vector are extracted through the Gabor filter.
  • the identity portrait information includes an identity portrait in BMP format
  • the security portrait information includes a security portrait in BMP format.
  • I(x,y) is the gray distribution of an image, and its Gabor feature is expressed as:
  • ⁇ ⁇ , v (x, y) is the Gabor kernel function
  • O ⁇ , v (x, y) is the Gabor wavelet convolution result corresponding to the scale ⁇ and direction ⁇ .
  • ⁇ ⁇ , v (x, y) is expressed as:
  • the resulting feature vector X is the identity feature vector; when processing the security portrait, the resulting feature vector X'is the security feature vector.
  • Step 306 Calculate the similarity percentage between the security feature vector and the identity feature vector.
  • the amplitude and phase of O k (z) are Ak and ⁇ k respectively , then Combining Ok (z) of different scales and directions constitutes the feature vector X of the image at the z position.
  • Step 308 Obtain a preset percentage threshold.
  • the preset percentage threshold is generally set at 60% to 95%.
  • Step 310 Compare the similarity percentage with a preset percentage threshold and obtain a portrait comparison result.
  • Gabor wavelet multi-resolution and multi-directional reflect the local characteristics of the image, so it is insensitive to the response to light, has better adaptability to light, and extracts image features more accurately.
  • acquiring the identification information and real-time security inspection image of the security inspection object specifically includes the following steps:
  • Step 402 Obtain identification information.
  • the terminal 102 obtains the ID number, passport number of the security check target ID card, and/or obtains the fingerprint of the security check target.
  • Step 404 Prompt for random action according to the obtained identity identification information.
  • the terminal 102 After confirming that the identification information is successfully obtained, the terminal 102 prompts a random action.
  • the random action prompt generally prompts the security subject to perform a certain facial operation. For example, the random action prompt may be "please smile” or "please blink”.
  • Step 406 Acquire continuous shooting images corresponding to the random action prompt.
  • the terminal 102 acquires continuous captured images corresponding to the random action.
  • Step 408 Extract the facial feature values in the continuously captured images and calculate the feature change value of the facial feature values.
  • the terminal 102 obtains the facial feature extraction position corresponding to the random action according to the random action prompt, and extracts the corresponding facial feature value in the continuously captured image according to the facial feature extraction position, and further calculates the feature change value of the facial feature value. For example, if the random action is "please blink", the terminal 102 obtains the eye feature extraction position according to the random action, and extracts the pixel value of the eye feature extraction position in the continuously captured image and the pixel value in the vicinity (ie, facial feature value ), calculate the difference between the pixel of the eye feature extraction position and the neighboring pixel according to the local binary mode (that is, the feature change value).
  • Step 410 Obtain a preset facial change threshold, and compare the feature change value with the preset facial change threshold.
  • the terminal 102 obtains the preset facial change threshold according to the random action prompt, and compares the feature change value with the preset facial change threshold.
  • step 412 when the feature change value is not less than the preset facial change threshold, a real-time security image is extracted from the continuously captured images.
  • the terminal 102 extracts a frame of images from the continuously captured images as a real-time security inspection image.
  • the terminal 102 may randomly extract images such as the first frame image, the Nth frame image, or the last frame image as real-time security inspection images.
  • the terminal 102 may use the last frame of the image as a real-time security inspection image.
  • image features are extracted from the real-time security inspection image, and the micro-expression similarity is obtained according to the image features and the detailed features of the preset micro-expression, specifically including the following steps:
  • Step 502 Extract image features from the real-time security inspection image.
  • micro-expression database There are more than 200,000 detail features in the micro-expression database. The same micro-expression will greatly change according to different people's behavior and habits. In order to achieve the accurate definition of micro-expressions, multiple detail features are used to reflect the same type of micro-expressions. In one embodiment, there are 54 preset micro-expressions, each type of micro-expression has 1000-4000 preset detail features, but each preset detail feature has a unique feature parameter.
  • the terminal extracts the existing image features from the real-time security inspection image according to the feature parameters of the preset detailed features in the micro-expression database.
  • Step 504 Compare the image feature with each preset detail feature to obtain a feature comparison result.
  • Each image feature and preset detail feature have unique feature parameters.
  • the terminal compares the image feature with the feature parameters of all preset detail features one by one, determines whether the feature parameters of the two are consistent, and then generates a feature comparison To the result. For example, in the real-time security inspection image, there is an image feature of “the arc at the corner of the mouth is 3 degrees”, and the feature parameter of the image feature is “radian is 3 degrees”.
  • the feature parameter of the image feature is consistent with the feature parameter of the preset detail feature "the arc of the mouth angle is 3 degrees”, and the feature parameter of the preset detail feature "the arc of the mouth angle is 5 degrees” is inconsistent.
  • Step 506 When it is determined that the image feature and the detail feature are the same feature according to the feature comparison result, search for the micro-expression category corresponding to each extracted image feature.
  • the terminal When judging that the image feature and the detail feature are the same feature according to the feature comparison result, the terminal performs micro-expression classification attribute search in turn according to each detail feature that is the same as the image feature, and determines the micro-expression category corresponding to the image feature.
  • the detail feature of the preset micro-expression is that the arc of the corner of the mouth is 3 degrees
  • the micro-expression category of the detail feature is "mouth corner" "Upward”
  • the classification of the micro-expression corresponding to the image feature is "uplift of the corner of the mouth”.
  • Step 508 Calculate the proportion of the number of image features corresponding to each micro-expression category in the total number of extracted image features to obtain the similarity of micro-expressions under each micro-expression category.
  • the terminal calculates the number of image features in each micro-expression category, and calculates the proportion of the number of image features in each micro-expression category in the total number of extracted image features.
  • the proportion is the similarity of the micro-expressions of each micro-expression.
  • calculating the emotion value of the security inspection object in each emotion mode according to the expression coefficient and the appearance probability includes: obtaining a preset number of extractions; sorting the micro expressions according to the corresponding appearance probability from large to small, and sorting
  • the top micro-expressions are extracted as micro-expressions to be processed, and the number of extracted micro-expressions is the same as the preset extraction amount; the emotions of the security object under each emotion mode are calculated according to the expression coefficients and the probability of appearance of the micro-expressions to be processed Value.
  • a preset number of extractions can be set, generally 4-10.
  • the terminal accumulates the product of the extracted expression coefficient of the micro-expression to be processed and the appearance probability to obtain the emotion value of the security inspection object in each emotion mode.
  • the method further includes: obtaining an emotion abnormality threshold range corresponding to the emotion pattern; comparing the emotion value of the security inspection object with the emotion abnormality threshold range; when the emotion value When it is within the range of the emotional abnormality threshold, confirmation information of emotional abnormality is generated.
  • Each emotion has a corresponding range of emotion abnormality thresholds, and the range of emotion abnormality thresholds of different emotions may not be the same.
  • the emotion value of the security inspection object is generally within the range of the emotional abnormality threshold, so the emotional abnormality threshold range is set, and by judging the security value of the security inspection object, it can be estimated whether the security inspection object exists Measures that endanger social security.
  • the terminal can determine whether the current emotion of the security inspection object is abnormal according to the emotion value.
  • the terminal When the security inspection object is judged to be emotionally abnormal, the terminal generates emotional abnormality confirmation information and displays it to the smart terminal held by the police or security inspector, so that the police or security inspector can understand the status of the security inspection object and communicate with the security inspection object.
  • the confirmation information of the emotion abnormality may be "Warning!
  • the current emotion mode of the security inspection object whose ID number is "XXXXXX” is stress, and the person's emotion threshold falls into the stress anomaly Threshold range, this person's emotions are too nervous, please communicate with the security inspection object to confirm!.
  • steps in the flowcharts of FIGS. 2-5 are displayed in order according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless clearly stated in this article, the execution of these steps is not strictly limited in order, and these steps may be executed in other orders. Moreover, at least some of the steps in FIGS. 2-5 may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but may be executed at different times. These sub-steps or stages The execution order of is not necessarily sequential, but may be executed in turn or alternately with at least a part of other steps or sub-steps or stages of other steps.
  • a security inspection auxiliary analysis device including: a security inspection acquisition module 602, an identity image search module 604, a portrait comparison module 606, a micro expression analysis module 608, and an expression coefficient acquisition module 610.
  • a calculation module 612 and an emotion pattern determination module 614 where:
  • the security inspection obtaining module 602 is used to obtain the identification information and real-time security inspection images of the security inspection object.
  • the identity image search module 604 is used to search for the identity image corresponding to the identification information.
  • the portrait comparison module 606 is used to perform portrait comparison between the real-time security inspection image and the identity image to obtain a portrait comparison result.
  • the micro-expression analysis module 608 is used to extract the image features from the real-time security image when determining that the portrait corresponding to the real-time security image and the portrait corresponding to the identity image belong to the same person according to the portrait comparison result, and to compare the image characteristics with the preset micro-expression The detailed features are compared to obtain the similarity of micro-expressions.
  • the expression coefficient acquisition module 610 is used to search for the expression coefficient corresponding to each preset micro expression in each emotion mode.
  • the calculation module 612 is used to calculate the emotion value of the security inspection object in each emotion mode according to the expression coefficient and the micro-expression similarity.
  • the emotion pattern determination module 614 is used to determine the emotion pattern of the security inspection object according to the emotion value.
  • the portrait comparison module includes a portrait extraction unit, a feature vector extraction unit, a similarity calculation unit, a percentage threshold acquisition unit, and a portrait comparison unit, wherein:
  • the portrait extraction unit is used to extract the security inspection portrait information from the real-time security inspection image, and extract the identity portrait information from the identity image.
  • the feature vector extraction unit is used to extract the identity feature vector from the identity portrait information and the security detection feature vector from the security inspection portrait information.
  • the similarity calculation unit is used to calculate the similarity percentage of the security feature vector and the identity feature vector.
  • the percentage threshold obtaining unit is used to obtain a preset percentage threshold.
  • the portrait comparison unit is used to compare the similarity percentage with a preset percentage threshold and obtain a portrait comparison result.
  • the security inspection acquisition module includes an identity recognition acquisition unit, an action prompt unit, a captured image acquisition unit, a feature change value calculation unit, a facial change threshold acquisition comparison unit, and a security inspection image extraction unit, wherein:
  • the identity recognition obtaining unit is used for obtaining identity recognition information.
  • the action prompting unit is used for prompting a random action according to the acquired identification information.
  • the captured image acquisition unit is used to acquire a continuous captured image corresponding to a random action prompt.
  • the feature change value calculation unit is used to extract the facial feature values in the continuously captured images and calculate the feature change values of the facial feature values.
  • the face change threshold acquisition and comparison unit is used to obtain a preset face change threshold and compare the feature change value with the preset face change threshold.
  • the security image extraction unit is used to extract real-time security images from the continuously captured images when the feature change value is not less than the preset facial change threshold.
  • the micro expression analysis module includes a feature extraction unit, a feature comparison unit, an attribute search unit, a classification unit, and a probability calculation unit, where:
  • a feature extraction unit is used to extract image features from the real-time security inspection image.
  • the feature comparison unit is configured to compare the image feature with each preset detailed feature to obtain a feature comparison result.
  • the attribute searching unit is configured to search for the micro-expression category corresponding to each of the extracted detailed features when it is determined that the image feature and the preset detailed feature are the same feature according to the feature comparison result.
  • a similarity calculation unit used to calculate the proportion of the number of image features corresponding to each micro-expression category in the total number of extracted image features, to obtain the micro-expression similarity under each micro-expression category .
  • the calculation module includes a number extraction unit, a micro-expression extraction unit, and an emotion value calculation unit, where:
  • the number extraction unit is used to obtain the preset extraction number.
  • Micro-expression extraction unit used to sort the micro-expressions according to the corresponding appearance probability from large to small, extract the top micro-expressions as micro-expressions to be processed, and extract the number and preset of micro-expressions to be processed The extraction quantity is consistent.
  • the emotion value calculation unit is used to calculate the emotion value of the security inspection object in each emotion mode according to the expression coefficient and the appearance probability of the micro-expression to be processed.
  • the security inspection auxiliary analysis device further includes an abnormality threshold acquisition module, an emotion value comparison module, and an abnormality information generation module, wherein:
  • the abnormal threshold acquisition module is used to acquire an emotional abnormal threshold range corresponding to the emotional pattern.
  • the emotion value comparison module is used to compare the emotion value of the security inspection object with the emotion abnormality threshold range.
  • the abnormal information generating module is used to generate emotional abnormality confirmation information when the emotional value is within the range of the emotional abnormality threshold.
  • Each module in the aforementioned security inspection auxiliary analysis device may be implemented in whole or in part by software, hardware, or a combination thereof.
  • the above modules may be embedded in the hardware form or independent of the processor in the computer device, or may be stored in the memory in the computer device in the form of software so that the processor can call and execute the operations corresponding to the above modules.
  • a computer device is provided.
  • the computer device may be a terminal, and an internal structure diagram thereof may be as shown in FIG. 7.
  • the computer equipment includes a processor, a memory, a network interface, a display screen, and an input device connected through a system bus.
  • the processor of the computer device is used to provide computing and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system and computer-readable instructions.
  • the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium.
  • the network interface of the computer device is used to communicate with external terminals through a network connection.
  • the display screen of the computer device may be a liquid crystal display screen or an electronic ink display screen
  • the input device of the computer device may be a touch layer covered on the display screen, or may be a button, a trackball, or a touchpad provided on the computer device housing , Can also be an external keyboard, touchpad or mouse.
  • FIG. 7 is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer equipment to which the solution of the present application is applied.
  • the specific computer equipment may It includes more or fewer components than shown in the figure, or some components are combined, or have a different component arrangement.
  • a computer device includes a memory and one or more processors.
  • the memory stores computer-readable instructions.
  • the one or more processors perform the following steps:
  • the image features are extracted from the real-time security image, and the micro-expression similarity is obtained according to the image features and the detailed features of the preset micro-expression;
  • the expression coefficient and the similarity of micro-expressions calculate the emotion value of the security inspection object under each emotion mode; determine the emotion mode of the security inspection object according to the emotion value.
  • the processor when performing the step of comparing the real-time security image with the identity image to obtain a portrait comparison result, it is also used to: from the real-time security image Extract the security portrait information, and extract the identity portrait information from the identity image; extract the identity feature vector from the identity portrait information, extract the security feature vector from the security portrait information; calculate the percentage of similarity between the security feature vector and the identity feature vector; Set a percentage threshold; and compare the percentage of similarity with the preset percentage threshold and get a portrait comparison result.
  • the processor when performing the step of acquiring the identification information and real-time security image of the security inspection object, it is also used to: obtain identification information; perform randomization based on the obtained identification information Action prompts; obtain continuous captured images corresponding to random action prompts; extract facial feature values in continuous captured images and calculate feature change values of facial feature values; obtain preset facial change thresholds, and perform feature change values with preset facial change thresholds Comparison; and when the feature change value is not less than the preset facial change threshold, real-time security inspection images are extracted from the continuously captured images.
  • the step of extracting image features from the real-time security image and obtaining the micro-expression similarity according to the detailed features of the image features and preset micro-expressions It is also used for: extracting image features from real-time security inspection images; comparing the image features with each preset detail feature to obtain a feature comparison result; when judging that the image feature and the preset detail feature are the same feature based on the feature comparison result, Find the micro-expression categories corresponding to the extracted detailed features; and calculate the proportion of the number of image features corresponding to each micro-expression category in the total number of extracted image features to obtain the micro-expression similarity under each micro-expression category .
  • the processor executes the computer-readable instruction, after performing the step of determining the emotion pattern of the security inspection object according to the emotion value, it is further used to: obtain an emotion abnormality threshold range corresponding to the emotion pattern; The emotion value of the security inspection object is compared with the range of the emotion abnormality threshold; and when the emotion value is within the range of the emotion abnormality threshold, confirmation information of the emotion abnormality is generated.
  • One or more non-volatile computer-readable storage media storing computer-readable instructions.
  • the computer-readable instructions When executed by one or more processors, the one or more processors perform the following steps:
  • the image features are extracted from the real-time security image, and the micro-expression similarity is obtained according to the image features and the detailed features of the preset micro-expression;
  • Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory can include random access memory (RAM) or external cache memory.
  • RAM random access memory
  • DRAM dynamic RAM
  • SDRAM synchronous DRAM
  • DDRSDRAM double data rate SDRAM
  • ESDRAM enhanced SDRAM
  • SLDRAM synchronous chain (Synchlink) DRAM
  • RDRAM direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Landscapes

  • Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Theoretical Computer Science (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Databases & Information Systems (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Software Systems (AREA)
  • Health & Medical Sciences (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Image Analysis (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)

Abstract

一种安检辅助分析方法，包括：获取安检对象的身份识别信息和实时安检图像；查找与身份识别信息对应的身份图像；将实时安检图像与身份图像进行人像比对得到人像比对结果；当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同一人时，从实时安检图像中提取图像特征，并将图像特征与预设微表情的细节特征进行比对得到微表情相似度；查找各情绪模式下各预设微表情对应的表情系数；根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值；根据情绪数值确定安检对象的情绪模式。

Description

安检辅助分析方法、装置、计算机设备和存储介质
相关申请的交叉引用
本申请要求于2018年12月18日提交中国专利局，申请号为2018115493067，申请名称为“安检辅助分析方法、装置、计算机设备和存储介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及一种安检辅助分析方法、装置、计算机设备和存储介质。
背景技术
现有的安检设备主要是检查人们随身携带的行李物品或对人体进行全身检查，检查人体是否携带违禁品，但这些都不能通过犯罪分子的特征主动识别罪犯。在现有安检模式中，安检员只能在显示屏上看到物品的X光图像、当前柜台号等信息，无法全面了解安检现场的情况，只能根据片面信息进行安检，可能存在漏检或者错检等现象，导致安检的准确率低，安检效果差，且对于危险的防范比较被动。
发明内容
根据本申请公开的各种实施例，提供一种安检辅助分析方法、装置、计算机设备和存储介质。
一种安检辅助分析方法，包括：
获取安检对象的身份识别信息和实时安检图像；
查找与所述身份识别信息对应的身份图像；
将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的人像同属一人时，从所述实时安检图像中提取图像特征，并将所述图像特征与预设微表情的细节特征进行相似度比对，得到微表情相似度；
查找各情绪模式下各所述预设微表情对应的表情系数；
根据所述表情系数及所述微表情相似度计算各所述情绪模式下所述安检对象的情绪数值；及
根据所述情绪数值确定所述安检对象的情绪模式。
一种安检辅助分析装置，包括：
安检获取模块，用于获取所述安检对象的身份识别信息和实时安检图像；
身份图像查找模块，用于查找与所述身份识别信息对应的身份图像；
人像比对模块，用于将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
微表情分析模块，用于当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的人像同属一人时，从所述实时安检图像中提取图像特征，并根据所述图像特征与预设微表情的细节特征得到微表情相似度；
表情系数获取模块，用于查找各情绪模式下各所述预设微表情对应的表情系数；
计算模块，用于根据所述表情系数及所述微表情相似度计算各情绪模式下所述安检对象的情绪数值；及
情绪模式判定模块，用于根据所述情绪数值确定所述安检对象的情绪模式。
一种计算机设备，包括存储器和一个或多个处理器，存储器中存储有计算机可读指令，计算机可读指令被处理器执行时实现本申请任意一个实施例中提供的安检辅助分析方法的步骤。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器实现本申请任意一个实施例中提供的安检辅助分析方法的步骤。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征和优点将从说明书、附图以及权利要求书变得明显。
附图说明
为了更清楚地说明本申请实施例中的技术方案，下面将对实施例中所需要使用的附图作简单地介绍，显而易见地，下面描述中的附图仅仅是本申请的一些实施例，对于本领域普通技术人员来讲，在不付出创造性劳动的前提下，还可以根据这些附图获得其它的附图。
图1为根据一个或多个实施例中安检辅助分析方法的应用场景图。
图2为根据一个或多个实施例中安检辅助分析方法的流程示意图。
图3为根据一个或多个实施例中人像比对步骤的流程示意图。
图4为根据一个或多个实施例中安检信息获取步骤的流程示意图。
图5为根据一个或多个实施例中微表情分类步骤的流程示意图。
图6为根据一个或多个实施例中安检辅助分析装置的框图。
图7为根据一个或多个实施例中计算机设备的框图。
具体实施方式
为了使本申请的技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请提供的安检辅助分析方法，可以应用于如图1所示的应用环境中。终端102通过网络与服务器104进行通信。终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式安检智能设备，服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。在其中一个实施例中，终端102为设置在火车站或者机场等人流量较大位置处的安检辅助分析装置。安检辅助分析装置既可以通过获取设置在外部的摄像头间接获取安检对象的实时安检图像，也可以采用摄像模块对安检对象进行实时拍摄直接获取实时安检图像。终端102将获取到的身份识别信息上传给服务器104，服务器104根据接收到的身份识别信息查找身份图像，并将对应的身份图像发送给终端102，终端102接收到身份图像后，先将实时安检图像与身份图像进行人像比对得到人像比对结果；而后进行微表情分析。服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现，为终端102提供数据接入服务。
在一些实施例中，如图2所示，提供了一种安检辅助分析方法，以该方法应用于图1中的终端102为例进行说明，包括以下步骤：
步骤202，获取安检对象的身份识别信息和实时安检图像。
在本实施例中，安检对象可以是乘客或者乘务人员。身份识别信息可以是安检对象事先在安检系统中存储的身份证号码、护照号码、指纹等等。终端102可以获取安检对象身份证的身份证号码和/或获取安检对象的指纹。
终端102在获取安检对象的身份识别信息的同时，获取实时安检图像。实时安检图像是安检对象在经过预定位置处时的拍摄图像。拍摄图像可以根据需要为一至多张。当拍摄图像仅为一张时，该拍摄图像就是实时安检图像；当拍摄图像为多张时，可以设定其中某一张拍摄图像为实时安检图像，也可以设定某几张拍摄图像均为实时安检图像。
步骤204，查找与身份识别信息对应的身份图像。
具体地，终端102将获取的身份识别信息发送给服务器，并在服务器内根据身份识别信息进行查找，并获取对应的身份图像。在一实施例中，终端102可以生成一个含有身份识别信息的图像获取请求发送给服务器，并接收服务器基于该图像获取请求返回被请求的身份图像。服务器根据图像获取请求中身份识别信息进行查找，并将查找到的身份图像发送给终端102。
步骤206，将实时安检图像与身份图像进行人像比对得到人像比对结果。
判断实时安检图像对应的人像与身份图像对应的人像是否同属一人，可以采用人脸图像比对识别方法也可以采用两照比对识别方法将实时安检图像与身份图像进行人像比对判断。
具体地，当采用人脸图像比对识别方法时，终端102先对实时安检图像和身份图像进行归一化处理；再获取事先建立好的通用三维人脸模型库，通过预先训练好的多任务学习神经网络对实时安检图像和身份图像进行人脸检测及人脸关键点信息提取确定各器官的位置，进而对实时安检图像和身份图像中的人脸头像提取对应的尺寸数据；根据尺寸数 据计算人像中的脸部比例数据；判断身份头像和安检头像的脸部比例数据是否一致并生成人像比对结果。
通用三维人脸模型库适用于所有人群，存储有将人脸分为眼眉、眼睛、鼻子、嘴巴、下巴、脸颊、额头、耳朵以及后脑共九个区域分类以及分类数据。每个区域分类分别定义了区域轮廓点以及特征点。分类数据还记录了区域轮廓点以及特征点的几何关系。因而，采用通用三维人脸模型库可以迅速定位出眼眉、眼睛、鼻子、嘴巴、下巴、脸颊、额头以及耳朵在人脸的位置，进而提取出各器官在人脸的尺寸数据。终端102根据尺寸数据分别计算实时安检图像和身份图像中人像的脸部比例数据，并将两种图像的脸部比例数据进行比对判断。当脸部比例数据一致时，生成实时安检图像对应的人像与身份图像对应的人像同属一人的人像比对结果；当脸部比例数据不一致时，生成实时安检图像对应的人像与身份图像对应的人像不是同一人的人像比对结果。
步骤208，当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同属一人时，从实时安检图像中提取图像特征，并根据图像特征与预设微表情的细节特征得到微表情相似度。
当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同属一人时，终端102从实时安检图像中提取图像特征，并将图像特征与预先存储的预设微表情的细节特征进行相似度比对，得到微表情相似度。例如，将实时安检图像中提取出的图像特征与事先存储的微表情“眯眼睛”、“嘴角上扬”、“嘴角下拉”、“眨眼睛”以及“露牙齿”的图像特征逐个进行相似度比对，比对数值分别为40％、80％、0％、10％以及90％。因此，实时安检图像中仅存在微表情“眯眼睛”、“嘴角上扬”、“眨眼睛”以及“露牙齿”，且微表情相似度分别为40％、80％、10％以及90％。
步骤210，查找各情绪模式下各微表情对应的表情系数。
终端根据微表情分别查找对应的各情绪模式下的表情系数。在一些实施例中，情绪模式有7种，分别为高兴、伤心、害怕、愤怒、厌恶、惊讶以及轻蔑。但是不同微表情与各情绪模式之间的相关性不一样，所以表情系数也不一样，例如，微表情“嘴角上扬”与高兴、伤心、害怕、愤怒、厌恶、惊讶以及轻蔑对应的表情系数分别为9、0、0、5、3、1以及2。
步骤212，根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值。
终端根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值，例如，在情绪模式为高兴的情况下，提取到的微表情为“眯眼睛”、“嘴角上扬”、“眨眼睛”以及“露牙齿”，微表情相似度分别为40％、80％、10％以及90％，微表情“眯眼睛”与高兴的表情系数为4，微表情“嘴角上扬”与高兴的表情系数为9，微表情“眨眼睛”与高兴的表情系数为1，微表情“露牙齿”与高兴的表情系数为10，此时，高兴情绪模式下安检对象的情绪数值为17.9(＝40％*4+80％*9+10％*1+90％*10)。
步骤214，根据情绪数值确定安检对象的情绪模式。
具体地，终端根据各情绪模式的情绪数值将情绪数值最大的情绪模式设定为安检对象的情绪模式。
上述安检辅助分析方法中，先获取安检对象的身份图像和实时安检图像，对两个图像进行比对分析，确定安检对象的身份，从而可以对安检对象的行踪进行追溯，而且也可以降低冒用他人身份的发生概率。而后在确认安检对象身份的情况下对安检对象的面部表情进行微表情分析，得到安检对象的当前情绪模式，从而确定安检对象的实时情绪，进而实现对安检对象的进一步了解，辅助进行安检决策，提高安检的准确率和安检效果。
在一些实施例中，如图3所示，将实时安检图像与身份图像进行人像比对得到人像比对结果，具体包括以下步骤：
步骤302，从实时安检图像中提取安检人像信息，并从身份图像中提取身份人像信息。
终端对实时安检图像进行人像检测，确定人像在实时安检图像的位置，而后根据确定的图像位置从实时安检图像中提取安检人像信息。同样的，终端对身份图像也进行人像检测，确定人像在身份图像的位置，而后根据确定的图像位置从身份图像中提取身份人像信息。
步骤304，从身份人像信息中提取身份特征矢量，从安检人像信息中提取安检特征矢量。
每个人的身高、体重、肩宽等等尺寸比例都是独一无二的，因而采用对不同人的人像信息进行分析时，得到的图像特征矢量也不可能完全一致，所以可以用来判断是否是同一个人。在其中一个实施例中，身份特征矢量和安检特征矢量均是通过Gabor滤波器进行提取。身份人像信息包含含有BMP格式的身份人像，安检人像信息包含含有BMP格式的安检人像。
I(x,y)为一幅图像的灰度分布，其Gabor特征表示为：
O μ，v(x,y)＝I(x,y)*ψ μ，v(x,y)，也可以表示为O k(z)＝I(z)*ψ(k,z)，
其中，ψ μ，v(x,y)是Gabor核函数，O μ，v(x,y)就是相应于尺度μ、方向ν的Gabor小波卷积结果。ψ μ，v(x,y)表示为下式：
Figure PCTCN2019071257-appb-000001
其中，
Figure PCTCN2019071257-appb-000002
确定了Gabor内核的尺度和方向，z＝(x,y)为空间位置坐标，k ν＝k max/f ν为采样尺度，φ μ＝πμ/8为采样方向，k max为最大频率，f是频域中的内核间隔因子。
对于人脸识别而言，一般选用5个尺度8个方向的Gabor小波卷积结果，即v∈{0，...,4},μ∈{0，...,7}，其余参数设为k max＝π/2，
Figure PCTCN2019071257-appb-000003
以及σ＝2π。由于直接将所有尺度、方向的Gabor小波卷积结果变换表示成一个级联的特征矢量，会导致图像的维数太高而难于处理，因而对各Gabor小波卷积结果进行下采样处理，采样因子为ρ，所以一幅图像的级联的特征矢量X为：
Figure PCTCN2019071257-appb-000004
当终端对身份人像进行处理，得到的特征矢量X就是身份特征矢量；对安检人像进行处理，得到的特征矢量X’就是安检特征矢量。
步骤306，计算安检特征矢量与身份特征矢量的相似度百分比。
O k(z)的幅值和相位分别为A k和Ф k，则
Figure PCTCN2019071257-appb-000005
组合不同尺度和方向的O k(z)构成图像在z位置处的特征矢量X。
身份特征矢量X和安检特征矢量X’的相似性S A计算公式如下：
Figure PCTCN2019071257-appb-000006
终端将在步骤304中计算得到的身份特征矢量X和安检特征矢量X’代入相似性S A计算公式得到相似度百分比。
步骤308，获取预设百分比阈值。
考虑到拍摄环境、拍摄设备以及拍摄者的变化，设置一个预设百分比阈值。为了降低误判率，一般预设百分比阈值设置在60％～95％。
步骤310，将相似度百分比与预设百分比阈值进行比较并得到人像比对结果。
判断相似度百分比与预设百分比阈值之间的大小，当相似度百分比大于预设百分比阈值时，生成实时安检图像对应的人像与身份图像对应的人像同属一人的人像比对结果；当相似度百分比小于预设百分比阈值时，生成实时安检图像对应的人像与身份图像对应的人像不是同一人的人像比对结果。
在上述安检辅助分析方法中，由于Gabor小波多分辨率、多方向地反映图像局部特性，所以对于光照的反应不敏感，对光照具有更好的适应性，对图像特征的提取也更准确。
在其中一个实施例中，如图4所示，获取安检对象的身份识别信息和实时安检图像，具体包括以下步骤：
步骤402，获取身份识别信息。
终端102获取安检对象身份证的身份证号码、护照号码和/或获取安检对象的指纹。
步骤404，根据获取的身份识别信息进行随机动作提示。
终端102确认身份识别信息获取成功后，进行随机动作提示。随机动作提示一般提示安检对象进行某一面部操作，例如，随机动作提示可以是“请微笑”或者“请眨眼”等等。
步骤406，获取随机动作提示对应的连续拍摄图像。
在安检对象进行随机动作时，终端102获取随机动作对应的连续拍摄图像。
步骤408，提取连续拍摄图像中的面部特征值并计算面部特征值的特征变化数值。
终端102根据随机动作提示获取与随机动作对应的面部特征提取位置，并根据该面部特征提取位置提取连续拍摄图像中对应的面部特征值，进一步计算面部特征值的特征变化 数值。例如，随机动作为“请眨眼”，终端102根据该随机动作获取眼部特征提取位置，并提取连续拍摄图像中该眼部特征提取位置的像素值以及临近区域内的像素值(即面部特征值)，根据局部二值模式计算眼部特征提取位置的像素与邻域像素之间的差值(即特征变化数值)。
步骤410，获取预设面部变化阈值，将特征变化数值与预设面部变化阈值进行比较。
终端102根据随机动作提示获取预设面部变化阈值，并将特征变化数值与预设面部变化阈值的大小进行比较。
步骤412，当特征变化数值不小于预设面部变化阈值时，从连续拍摄图像中提取出实时安检图像。
当特征变化数值不小于预设面部变化阈值时，终端102从连续拍摄图像中提取出一帧图像作为实时安检图像。终端102可以随机提取如第一帧图像、第N帧图像或者最后一帧图像作为实时安检图像。优选地，因为在最后一帧图像中，安检对象已经做完提示动作，所以，为了减小对后续微表情分析的影响，终端102可以将最后一帧图像作为实时安检图像。
在上述安检辅助分析方法中，通过让安检对象进行随机动作，能够防范危险分子使用照片、3D人脸模型或者面具等方式进行验证，提高安检的安全系数。
在一些实施例中，如图5所示，从实时安检图像中提取图像特征，并根据图像特征与预设微表情的细节特征得到微表情相似度，具体包括以下步骤：
步骤502，从实时安检图像中提取图像特征。
微表情数据库中具有20多万个细节特征，同一个微表情根据不同人的行为习惯会有极大的变化，为了实现对微表情的准确定义采用多个细节特征对同一类微表情进行反映，在一实施例中，预设微表情共有54种，每类微表情具有1000～4000个预设细节特征，但每个预设细节特征具有唯一的特征参数。终端根据微表情数据库中预设细节特征的特征参数从实时安检图像中提取存在的图像特征。
步骤504，将所述图像特征与各预设细节特征进行比对得到特征比对结果。
每个图像特征和预设细节特征均具有唯一的特征参数，终端将图像特征与所有预设细节特征的特征参数一一进行比对，并判断两者的特征参数是否一致，然后生成特征比对结果。例如，实时安检图像中存在“嘴角向上的弧度为3度”图像特征，该图像特征的特征参数为“弧度为3度”。该图像特征的特征参数与预设细节特征“嘴角向上的弧度为3度”的特征参数是一致的，与预设细节特征“嘴角向上的弧度为5度”的特征参数是不一致的。
步骤506，当根据特征比对结果判断图像特征与细节特征为同一特征时，查找提取出的各图像特征对应的微表情类别。
当根据特征比对结果判断图像特征与细节特征为同一特征时，终端根据与图像特征相同的各细节特征依次进行微表情分类属性查找，并确定图像特征对应的微表情类别。在本 实施例中，预设微表情共有54种，微表情类别也存在54种，例如，预设微表情的细节特征为嘴角向上的弧度为3度，该细节特征的微表情类别为“嘴角上扬”，因而，当图像特征也为嘴角向上的弧度为3度时，图像特征对应的微表情分类就是“嘴角上扬”。
步骤508，计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征的总数中的所占比例，得到各所述微表情类别下的微表情相似度。
终端计算每个微表情类别下图像特征的数量，并计算各微表情类别下图像特征数量在提取出的图像特征总数中的所占比例，该所占比例就是各微表情的微表情相似度。
在其中一个实施例中，根据表情系数及出现概率计算各情绪模式下安检对象的情绪数值，包括：获取预设提取数量；将微表情根据对应的出现概率由大至小的顺序进行排序，将排在前列的微表情提取为待处理微表情，且提取出待处理微表情的数量与预设提取数量一致；根据待处理微表情的表情系数及出现概率计算得到各情绪模式下安检对象的情绪数值。
当终端获得的微表情数量比较多时，计算比较繁复，为了快速得到安检对象的情绪数值，可以设置预设提取数量，一般设置4～10个。终端将提取出的待处理微表情的表情系数和出现概率的乘积进行累加得到各情绪模式下安检对象的情绪数值。
在其中一个实施例中，根据情绪数值判定安检对象的情绪模式之后，方法还包括：获取情绪模式对应的情绪异常阈值范围；将安检对象的情绪数值与情绪异常阈值范围进行比对；当情绪数值处于情绪异常阈值范围内时，生成情绪异常确认信息。
每个情绪均有对应的情绪异常阈值范围，且不同情绪的情绪异常阈值范围不一定相同。当安检对象进行某些危害社会安全行为时，安检对象的情绪数值一般处于情绪异常阈值范围内，所以设定情绪异常阈值范围，通过对安检对象的情绪数值的判断，可以预估安检对象是否存在危害社会安全的举措。终端根据情绪数值可以确定安检对象的当前情绪是否异常。当安检对象被判定为情绪异常时，终端生成情绪异常确认信息并显示给协警或安检员持有的智能终端，以便协警或安检员了解安检对象的状况，并与安检对象进行沟通。例如，当安检对象情绪阈值落入紧张异常阈值范围时，情绪异常确认信息可以是“警告！身份证号为“XXXXXXX”的安检对象的当前情绪模式为紧张，此人的情绪阈值落入紧张异常阈值范围，此人情绪过于紧张，请与该名安检对象进行沟通确认！”。
应该理解的是，虽然图2-5的流程图中的各个步骤按照箭头的指示依次显示，但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明，这些步骤的执行并没有严格的顺序限制，这些步骤可以以其它的顺序执行。而且，图2-5中的至少一部分步骤可以包括多个子步骤或者多个阶段，这些子步骤或者阶段并不必然是在同一时刻执行完成，而是可以在不同的时刻执行，这些子步骤或者阶段的执行顺序也不必然是依次进行，而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在一些实施例中，如图6所示，提供了一种安检辅助分析装置，包括：安检获取模 块602、身份图像查找模块604、人像比对模块606、微表情分析模块608、表情系数获取模块610、计算模块612和情绪模式判定模块614，其中：
安检获取模块602，用于获取安检对象的身份识别信息和实时安检图像。
身份图像查找模块604，用于查找与身份识别信息对应的身份图像。
人像比对模块606，用于将实时安检图像与身份图像进行人像比对得到人像比对结果。
微表情分析模块608，用于当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同属一人时，从实时安检图像中提取图像特征，并将图像特征与预设微表情的细节特征进行比对，得到微表情相似度。
表情系数获取模块610，用于查找各情绪模式下各预设微表情对应的表情系数。
计算模块612，用于根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值。
情绪模式判定模块614，用于根据情绪数值确定安检对象的情绪模式。
在一些实施例中，人像比对模块包括人像提取单元、特征矢量提取单元、相似度计算单元、百分比阈值获取单元以及人像比对单元，其中：
人像提取单元，用于从实时安检图像中提取安检人像信息，并从身份图像中提取身份人像信息。
特征矢量提取单元，用于从身份人像信息中提取身份特征矢量，从安检人像信息中提取安检特征矢量。
相似度计算单元，用于计算安检特征矢量与身份特征矢量的相似度百分比。
百分比阈值获取单元，用于获取预设百分比阈值。
人像比对单元，用于将相似度百分比与预设百分比阈值进行比较并得到人像比对结果。
在其中一个实施例中，安检获取模块包括身份识别获取单元、动作提示单元、拍摄图像获取单元、特征变化值计算单元、面部变化阈值获取比较单元以及安检图像提取单元，其中：
身份识别获取单元，用于获取身份识别信息。
动作提示单元，用于根据获取的身份识别信息进行随机动作提示。
拍摄图像获取单元，用于获取随机动作提示对应的连续拍摄图像。
特征变化值计算单元，用于提取连续拍摄图像中的面部特征值并计算面部特征值的特征变化数值。
面部变化阈值获取比较单元，用于获取预设面部变化阈值，将特征变化数值与预设面部变化阈值进行比较。
安检图像提取单元，用于当特征变化数值不小于预设面部变化阈值时，从连续拍摄图像中提取出实时安检图像。
在其中一个实施例中，微表情分析模块包括特征提取单元、特征比对单元、属性查找单元、分类单元以及概率计算单元，其中：
特征提取单元，用于从所述实时安检图像中提取图像特征。
特征比对单元，用于将所述图像特征与各预设细节特征进行比对得到特征比对结果。
属性查找单元，用于当根据所述特征比对结果判断所述图像特征与所述预设细节特征为同一特征时，查找提取出的各所述细节特征对应的微表情类别。
相似度计算单元，用于计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征总数中的所占比例，得到各所述微表情类别下的微表情相似度。
在一些实施例中，计算模块包括数量提取单元，微表情提取单元以及情绪数值计算单元，其中：
数量提取单元，用于获取预设提取数量。
微表情提取单元，用于将微表情根据对应的出现概率由大至小的顺序进行排序，将排在前列的微表情提取为待处理微表情，且提取出待处理微表情的数量与预设提取数量一致。
情绪数值计算单元，用于根据待处理微表情的表情系数及出现概率计算得到各情绪模式下安检对象的情绪数值。
在一些实施例中，安检辅助分析装置还包括异常阈值获取模块、情绪数值比对模块以及异常信息生成模块，其中：
异常阈值获取模块，用于获取情绪模式对应的情绪异常阈值范围。
情绪数值比对模块，用于将安检对象的情绪数值与情绪异常阈值范围进行比对。
异常信息生成模块，用于当情绪数值处于情绪异常阈值范围内时，生成情绪异常确认信息。
关于安检辅助分析装置的具体限定可以参见上文中对于安检辅助分析方法的限定，在此不再赘述。上述安检辅助分析装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中，也可以以软件形式存储于计算机设备中的存储器中，以便于处理器调用执行以上各个模块对应的操作。
在一些实施例中，提供了一种计算机设备，该计算机设备可以是终端，其内部结构图可以如图7所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口、显示屏和输入装置。其中，该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统和计算机可读指令。该内存储器为非易失性存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令被处理器执行时以实现一种安检辅助分析方法。该计算机设备的显示屏可以是液晶显示屏或者电子墨水显示屏，该计算机设备的输入装置可以是显示屏上覆盖的触摸层，也可以是 计算机设备外壳上设置的按键、轨迹球或触控板，还可以是外接的键盘、触控板或鼠标等。
本领域技术人员可以理解，图7中示出的结构，仅仅是与本申请方案相关的部分结构的框图，并不构成对本申请方案所应用于其上的计算机设备的限定，具体的计算机设备可以包括比图中所示更多或更少的部件，或者组合某些部件，或者具有不同的部件布置。
一种计算机设备，包括存储器和一个或多个处理器，存储器中储存有计算机可读指令，计算机可读指令被处理器执行时，使得一个或多个处理器执行以下步骤：
获取安检对象的身份识别信息和实时安检图像；
查找与身份识别信息对应的身份图像；
将实时安检图像与身份图像进行人像比对得到人像比对结果；
当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同属一人时，从实时安检图像中提取图像特征，并根据图像特征与预设微表情的细节特征得到微表情相似度；
查找各情绪模式下各预设微表情对应的表情系数；及
根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值；根据情绪数值确定安检对象的情绪模式。
在一些实施例中，处理器执行计算机可读指令时执行所述将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果的步骤时，还用于：从实时安检图像中提取安检人像信息，并从身份图像中提取身份人像信息；从身份人像信息中提取身份特征矢量，从安检人像信息中提取安检特征矢量；计算安检特征矢量与身份特征矢量的相似度百分比；获取预设百分比阈值；及将相似度百分比与预设百分比阈值进行比较并得到人像比对结果。
在其中一个实施例中，处理器执行计算机可读指令时执行所述获取安检对象的身份识别信息和实时安检图像的步骤时，还用于：获取身份识别信息；根据获取的身份识别信息进行随机动作提示；获取随机动作提示对应的连续拍摄图像；提取连续拍摄图像中的面部特征值并计算面部特征值的特征变化数值；获取预设面部变化阈值，将特征变化数值与预设面部变化阈值进行比较；及当特征变化数值不小于预设面部变化阈值时，从连续拍摄图像中提取出实时安检图像。
在一些实施例中，处理器执行计算机可读指令时执行所述从所述实时安检图像中提取图像特征，根据所述图像特征与预设微表情的细节特征得到微表情相似度的步骤时，还用于：从实时安检图像中提取图像特征；将图像特征与各预设细节特征进行比对得到特征比对结果；当根据特征比对结果判断图像特征与预设细节特征为同一特征时，查找提取出的各细节特征对应的微表情类别；及计算各微表情类别对应的图像特征的数量在提取出的图像特征的总数中的所占比例，得到各微表情类别下的微表情相似度。
在其中一个实施例中，处理器执行计算机可读指令时执行所述根据所述情绪数值判定所述安检对象的情绪模式的步骤之后，还用于：获取情绪模式对应的情绪异常阈值范围； 将安检对象的情绪数值与情绪异常阈值范围进行比对；及当情绪数值处于情绪异常阈值范围内时，生成情绪异常确认信息。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，计算机可读指令被一个或多个处理器执行时，使得一个或多个处理器执行以下步骤：：
获取安检对象的身份识别信息和实时安检图像；
查找与身份识别信息对应的身份图像；
将实时安检图像与身份图像进行人像比对得到人像比对结果；
当根据人像比对结果判定实时安检图像对应的人像与身份图像对应的人像同属一人时，从实时安检图像中提取图像特征，并根据图像特征与预设微表情的细节特征得到微表情相似度；
查找各情绪模式下各预设微表情对应的表情系数；
根据表情系数及微表情相似度计算各情绪模式下安检对象的情绪数值；及
根据情绪数值确定安检对象的情绪模式。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机可读指令来指令相关的硬件来完成，所述的计算机可读指令可存储于一非易失性计算机可读取存储介质中，该计算机可读指令在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合，为使描述简洁，未对上述实施例中的各个技术特征所有可能的组合都进行描述，然而，只要这些技术特征的组合不存在矛盾，都应当认为是本说明书记载的范围。
以上实施例仅表达了本申请的几种实施方式，其描述较为具体和详细，但并不能因此而理解为对发明专利范围的限制。应当指出的是，对于本领域的普通技术人员来说，在不脱离本申请构思的前提下，还可以做出若干变形和改进，这些都属于本申请的保护范围。因此，本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种安检辅助分析方法，包括：
    获取安检对象的身份识别信息和实时安检图像；
    查找与所述身份识别信息对应的身份图像；
    将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
    当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的人像同属一人时，从所述实时安检图像中提取图像特征，并根据所述图像特征与预设微表情的细节特征得到微表情相似度；
    查找各情绪模式下各所述预设微表情对应的表情系数；
    根据所述表情系数及所述微表情相似度计算各所述情绪模式下所述安检对象的情绪数值；及
    根据所述情绪数值确定所述安检对象的情绪模式。
  2. 根据权利要求1所述的方法，其特征在于，所述将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果，包括：
    从所述实时安检图像中提取安检人像信息，并从所述身份图像中提取身份人像信息；
    从所述身份人像信息中提取身份特征矢量，从所述安检人像信息中提取安检特征矢量；
    计算所述安检特征矢量与所述身份特征矢量的相似度百分比；
    获取预设百分比阈值；及
    将所述相似度百分比与所述预设百分比阈值进行比较并得到人像比对结果。
  3. 根据权利要求1所述的方法，其特征在于，所述获取安检对象的身份识别信息和实时安检图像，包括：
    获取所述身份识别信息；
    根据获取的所述身份识别信息进行随机动作提示；
    获取所述随机动作提示对应的连续拍摄图像；
    提取所述连续拍摄图像中的面部特征值并计算所述面部特征值的特征变化数值；
    获取预设面部变化阈值，将所述特征变化数值与所述预设面部变化阈值进行比较；及
    当所述特征变化数值不小于所述预设面部变化阈值时，从所述连续拍摄图像中提取出实时安检图像。
  4. 根据权利要求1所述的方法，其特征在于，所述从所述实时安检图像中提取图像特征，根据所述图像特征与预设微表情的细节特征得到微表情相似度，包括：
    从所述实时安检图像中提取图像特征；
    将所述图像特征与各预设细节特征进行比对得到特征比对结果；
    当根据所述特征比对结果判断所述图像特征与所述预设细节特征为同一特征时，查找提取出的各所述细节特征对应的微表情类别；及
    计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征的总数中的所占比例，得到各所述微表情类别下的微表情相似度。
  5. 根据权利要求1所述的方法，所述根据所述表情系数及所述微表情相似度计算各情绪模式下所述安检对象的情绪数值，包括：
    获取预设提取数量；
    将所述图像特征对应的微表情根据对应的所述微表情相似度由大至小的顺序进行排序，将排在前列的所述微表情提取为待处理微表情，且提取出所述待处理微表情的数量与所述预设提取数量一致；及
    根据所述待处理微表情的所述表情系数及所述微表情相似度计算得到各所述情绪模式下所述安检对象的情绪数值。
  6. 根据权利要求1至5任意一项所述的方法，其特征在于，所述根据所述情绪数值判定所述安检对象的情绪模式之后，包括：
    获取所述情绪模式对应的情绪异常阈值范围；
    将所述安检对象的情绪数值与所述情绪异常阈值范围进行比对；及
    当所述情绪数值处于所述情绪异常阈值范围内时，生成情绪异常确认信息。
  7. 一种安检辅助分析装置，其特征在于，所述装置包括：
    安检获取模块，用于获取所述安检对象的身份识别信息和实时安检图像；
    身份图像查找模块，用于查找与所述身份识别信息对应的身份图像；
    人像比对模块，用于将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
    微表情分析模块，用于当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的人像同属一人时，从所述实时安检图像中提取图像特征，并根据所述图像特征与预设微表情的细节特征得到微表情相似度；
    表情系数获取模块，用于查找各情绪模式下各所述预设微表情对应的表情系数；
    计算模块，用于根据所述表情系数及所述微表情相似度计算各情绪模式下所述安检对象的情绪数值；及
    情绪模式判定模块，用于根据所述情绪数值确定所述安检对象的情绪模式。
  8. 根据权利要求7所述的装置，其特征在于，所述微表情分析模块包括：
    特征提取单元，用于从所述实时安检图像中提取图像特征；
    特征比对单元，用于将所述图像特征与各预设细节特征进行比对得到特征比对结果；
    属性查找单元，用于当根据所述特征比对结果判断所述图像特征与所述预设细节特征为同一特征时，查找提取出的各所述细节特征对应的微表情类别；及
    相似度计算单元，用于计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征总数中的所占比例，得到各所述微表情类别下的微表情相似度。
  9. 根据权利要求7所述的装置，其特征在于，所述人像比对模块包括：
    人像提取单元，用于从所述实时安检图像中提取安检人像信息，并从所述身份图像中提取身份人像信息；
    特征矢量提取单元，用于从所述身份人像信息中提取身份特征矢量，从所述安检人像信息中提取安检特征矢量；
    相似度计算单元，用于计算所述安检特征矢量与所述身份特征矢量的相似度百分比；
    百分比阈值获取单元，用于获取预设百分比阈值；及
    人像比对单元，用于将所述相似度百分比与所述预设百分比阈值进行比较并得到人像比对结果。
  10. 根据权利要求7所述的装置，其特征在于，所述安检获取模块包括：
    身份识别获取单元，用于获取所述身份识别信息；
    动作提示单元，用于根据获取的所述身份识别信息进行随机动作提示；
    拍摄图像获取单元，用于获取所述随机动作提示对应的连续拍摄图像；
    特征变化值计算单元，用于提取所述连续拍摄图像中的面部特征值并计算所述面部特征值的特征变化数值；
    面部变化阈值获取比较单元，用于获取预设面部变化阈值，将所述特征变化数值与所述预设面部变化阈值进行比较；及
    安检图像提取单元，用于当所述特征变化数值不小于所述预设面部变化阈值时，从所述连续拍摄图像中提取出实时安检图像。
  11. 一种计算机设备，包括存储器及一个或多个处理器，所述存储器中储存有计算机可读指令，所述计算机可读指令被所述一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取安检对象的身份识别信息和实时安检图像；
    查找与所述身份识别信息对应的身份图像；
    将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
    当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的人像同属一人时，从所述实时安检图像中提取图像特征，并根据所述图像特征与预设微表情的细节特征得到微表情相似度；
    查找各情绪模式下各所述预设微表情对应的表情系数；
    根据所述表情系数及所述微表情相似度计算各所述情绪模式下所述安检对象的情绪数值；及
    根据所述情绪数值确定所述安检对象的情绪模式。
  12. 根据权利要求11所述的计算机设备，其特征在于，所述处理器执行所述计算机可读指令时执行所述将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果的步骤时，包括：
    从所述实时安检图像中提取安检人像信息，并从所述身份图像中提取身份人像信息；
    从所述身份人像信息中提取身份特征矢量，从所述安检人像信息中提取安检特征矢量；
    计算所述安检特征矢量与所述身份特征矢量的相似度百分比；
    获取预设百分比阈值；及
    将所述相似度百分比与所述预设百分比阈值进行比较并得到人像比对结果。
  13. 根据权利要求11所述的计算机设备，其特征在于，所述处理器执行计算机可读指令时执行所述获取安检对象的身份识别信息和实时安检图像的步骤时，包括：
    获取所述身份识别信息；
    根据获取的所述身份识别信息进行随机动作提示；获取所述随机动作提示对应的连续拍摄图像；
    提取所述连续拍摄图像中的面部特征值并计算所述面部特征值的特征变化数值；
    获取预设面部变化阈值，将所述特征变化数值与所述预设面部变化阈值进行比较；及
    当所述特征变化数值不小于所述预设面部变化阈值时，从所述连续拍摄图像中提取出实时安检图像。
  14. 根据权利要求11所述的计算机设备，其特征在于，所述处理器执行计算机可读指令时执行所述从所述实时安检图像中提取图像特征，根据所述图像特征与预设微表情的细节特征得到微表情相似度的步骤时，包括：
    从所述实时安检图像中提取图像特征；
    将所述图像特征与各预设细节特征进行比对得到特征比对结果；
    当根据所述特征比对结果判断所述图像特征与所述预设细节特征为同一特征时，查找提取出的各所述细节特征对应的微表情类别；及
    计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征的总数中的所占比例，得到各所述微表情类别下的微表情相似度。
  15. 根据权利要求11至14任意一项所述的计算机设备，其特征在于，所述处理器执行计算机可读指令时执行所述根据所述情绪数值判定所述安检对象的情绪模式的步骤之后，包括：
    获取所述情绪模式对应的情绪异常阈值范围；
    将所述安检对象的情绪数值与所述情绪异常阈值范围进行比对；及
    当所述情绪数值处于所述情绪异常阈值范围内时，生成情绪异常确认信息。
  16. 一个或多个存储有计算机可读指令的非易失性计算机可读存储介质，所述计算机可读指令被一个或多个处理器执行时，使得所述一个或多个处理器执行以下步骤：
    获取安检对象的身份识别信息和实时安检图像；
    查找与所述身份识别信息对应的身份图像；
    将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果；
    当根据所述人像比对结果判定所述实时安检图像对应的人像与所述身份图像对应的 人像同属一人时，从所述实时安检图像中提取图像特征，并根据所述图像特征与预设微表情的细节特征得到微表情相似度；
    查找各情绪模式下各所述预设微表情对应的表情系数；
    根据所述表情系数及所述微表情相似度计算各所述情绪模式下所述安检对象的情绪数值；及
    根据所述情绪数值确定所述安检对象的情绪模式。
  17. 根据权利要求16所述的存储介质，其特征在于，所述计算机可读指令被所述处理器执行所述将所述实时安检图像与所述身份图像进行人像比对得到人像比对结果的步骤时，包括：
    从所述实时安检图像中提取安检人像信息，并从所述身份图像中提取身份人像信息；
    从所述身份人像信息中提取身份特征矢量，从所述安检人像信息中提取安检特征矢量；
    计算所述安检特征矢量与所述身份特征矢量的相似度百分比；
    获取预设百分比阈值；及
    将所述相似度百分比与所述预设百分比阈值进行比较并得到人像比对结果。
  18. 根据权利要求16所述的存储介质，其特征在于，所述计算机可读指令被所述处理器执行所述获取安检对象的身份识别信息和实时安检图像的步骤时，包括：
    获取所述身份识别信息；
    根据获取的所述身份识别信息进行随机动作提示；获取所述随机动作提示对应的连续拍摄图像；
    提取所述连续拍摄图像中的面部特征值并计算所述面部特征值的特征变化数值；
    获取预设面部变化阈值，将所述特征变化数值与所述预设面部变化阈值进行比较；及
    当所述特征变化数值不小于所述预设面部变化阈值时，从所述连续拍摄图像中提取出实时安检图像。
  19. 根据权利要求16所述的存储介质，其特征在于，所述计算机可读指令被所述处理器执行所述从所述实时安检图像中提取图像特征，根据所述图像特征与预设微表情的细节特征得到微表情相似度的步骤时，包括：
    从所述实时安检图像中提取图像特征；
    将所述图像特征与各预设细节特征进行比对得到特征比对结果；
    当根据所述特征比对结果判断所述图像特征与所述预设细节特征为同一特征时，查找提取出的各所述细节特征对应的微表情类别；及
    计算各所述微表情类别对应的所述图像特征的数量在提取出的所述图像特征的总数中的所占比例，得到各所述微表情类别下的微表情相似度。
  20. 根据权利要求16至19任意一项所述的存储介质，其特征在于，所述计算机可读指令被所述处理器执行所述根据所述情绪数值判定所述安检对象的情绪模式的步骤之 后，包括：
    获取所述情绪模式对应的情绪异常阈值范围；
    将所述安检对象的情绪数值与所述情绪异常阈值范围进行比对；及
    当所述情绪数值处于所述情绪异常阈值范围内时，生成情绪异常确认信息。
PCT/CN2019/071257 2018-12-18 2019-01-11 安检辅助分析方法、装置、计算机设备和存储介质 Ceased WO2020124710A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811549306.7 2018-12-18
CN201811549306.7A CN109829362A (zh) 2018-12-18 2018-12-18 安检辅助分析方法、装置、计算机设备和存储介质

Publications (1)

Publication Number Publication Date
WO2020124710A1 true WO2020124710A1 (zh) 2020-06-25

Family

ID=66859773

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/071257 Ceased WO2020124710A1 (zh) 2018-12-18 2019-01-11 安检辅助分析方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN109829362A (zh)
WO (1) WO2020124710A1 (zh)

Cited By (10)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111898515A (zh) * 2020-07-24 2020-11-06 北京灵汐科技有限公司 一种行人身份识别的方法和装置
CN112220479A (zh) * 2020-09-04 2021-01-15 陈婉婷 基于遗传算法的被审讯单体情绪判断方法、装置和设备
CN112541384A (zh) * 2020-07-30 2021-03-23 深圳市商汤科技有限公司 对象查找方法及装置、电子设备及存储介质
CN112906660A (zh) * 2021-03-31 2021-06-04 浙江大华技术股份有限公司 安检预警方法和装置、存储介质及电子设备
CN114140542A (zh) * 2021-12-09 2022-03-04 成都智元汇信息技术股份有限公司 基于色彩补偿的切图方法、电子设备及介质
CN114255436A (zh) * 2020-09-11 2022-03-29 同方威视技术股份有限公司 基于人工智能的安检图像识别系统及其方法
CN115774031A (zh) * 2022-11-29 2023-03-10 同方威视科技江苏有限公司 查验任务处理方法、装置与电子设备
CN116244676A (zh) * 2023-05-09 2023-06-09 中航信移动科技有限公司 一种基于区块链的安全身份验证系统
CN117011921A (zh) * 2023-08-18 2023-11-07 德科信息有限公司 一种提高人脸识别率的图像处理方法及系统
CN119049159A (zh) * 2024-09-10 2024-11-29 重庆赛力斯凤凰智创科技有限公司 目标车门的解锁方法、装置、电子设备

Families Citing this family (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110580470A (zh) * 2019-09-12 2019-12-17 深圳壹账通智能科技有限公司 基于人脸识别的监测方法、装置、存储介质及计算机设备
CN111105530A (zh) * 2019-12-06 2020-05-05 国网山东省电力公司五莲县供电公司 一种基于专用通道感应的变电站门禁识别系统
CN111881822B (zh) * 2020-07-27 2024-09-17 芯算一体(深圳)科技有限公司 基于人脸识别的门禁方法、装置、设备及存储介质
CN112327268B (zh) * 2020-11-12 2024-12-27 奥谱毫芯(深圳)科技有限公司 基于毫米波信号的危险目标识别方法、装置、系统及介质
CN113139005B (zh) * 2021-04-22 2024-09-10 康键信息技术（深圳）有限公司 基于同人识别模型的同人识别方法及相关设备
CN114708524B (zh) * 2022-03-02 2025-04-25 浙江华视智检科技有限公司 一种包裹图像识别方法、装置、电子装置和存储介质
CN115035646B (zh) * 2022-06-07 2024-02-27 厦门民航凯亚有限公司 一种关于双感结合的防漏检方法

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104021397A (zh) * 2014-06-13 2014-09-03 中国民航信息网络股份有限公司 人脸识别比对方法及装置
CN106981140A (zh) * 2017-03-30 2017-07-25 广东微模式软件股份有限公司 一种电话卡自助服务一体化装置及其方法
CN108537160A (zh) * 2018-03-30 2018-09-14 平安科技（深圳）有限公司 基于微表情的风险识别方法、装置、设备及介质
CN208092811U (zh) * 2017-12-29 2018-11-13 同方威视技术股份有限公司 基于多生物特征的安检设备

Family Cites Families (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP6467965B2 (ja) * 2015-02-13 2019-02-13 オムロン株式会社 感情推定装置及び感情推定方法
CN105574478A (zh) * 2015-05-28 2016-05-11 宇龙计算机通信科技(深圳)有限公司 一种信息处理的方法和装置
CN106257489A (zh) * 2016-07-12 2016-12-28 乐视控股（北京）有限公司 表情识别方法及系统
CN107609458A (zh) * 2016-07-20 2018-01-19 平安科技（深圳）有限公司 基于人脸表情识别的情绪反馈方法及装置
CN108241836A (zh) * 2016-12-23 2018-07-03 同方威视技术股份有限公司 用于安检的方法及装置
CN107423707A (zh) * 2017-07-25 2017-12-01 深圳帕罗人工智能科技有限公司 一种基于复杂环境下的人脸情绪识别方法
CN107844748B (zh) * 2017-10-17 2019-02-05 平安科技（深圳）有限公司 身份验证方法、装置、存储介质和计算机设备
CN107895146B (zh) * 2017-11-01 2020-05-26 深圳市科迈爱康科技有限公司 微表情识别方法、装置、系统及计算机可读存储介质
CN108307037A (zh) * 2017-12-15 2018-07-20 努比亚技术有限公司 终端控制方法、终端及计算机可读存储介质
CN108764010A (zh) * 2018-03-23 2018-11-06 姜涵予 情绪状态确定方法及装置
CN108564007B (zh) * 2018-03-27 2021-10-22 深圳市智能机器人研究院 一种基于表情识别的情绪识别方法和装置

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104021397A (zh) * 2014-06-13 2014-09-03 中国民航信息网络股份有限公司 人脸识别比对方法及装置
CN106981140A (zh) * 2017-03-30 2017-07-25 广东微模式软件股份有限公司 一种电话卡自助服务一体化装置及其方法
CN208092811U (zh) * 2017-12-29 2018-11-13 同方威视技术股份有限公司 基于多生物特征的安检设备
CN108537160A (zh) * 2018-03-30 2018-09-14 平安科技（深圳）有限公司 基于微表情的风险识别方法、装置、设备及介质

Cited By (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111898515A (zh) * 2020-07-24 2020-11-06 北京灵汐科技有限公司 一种行人身份识别的方法和装置
CN112541384B (zh) * 2020-07-30 2023-04-28 深圳市商汤科技有限公司 可疑对象查找方法及装置、电子设备及存储介质
CN112541384A (zh) * 2020-07-30 2021-03-23 深圳市商汤科技有限公司 对象查找方法及装置、电子设备及存储介质
CN112220479A (zh) * 2020-09-04 2021-01-15 陈婉婷 基于遗传算法的被审讯单体情绪判断方法、装置和设备
CN114255436B (zh) * 2020-09-11 2024-03-19 同方威视技术股份有限公司 基于人工智能的安检图像识别系统及其方法
CN114255436A (zh) * 2020-09-11 2022-03-29 同方威视技术股份有限公司 基于人工智能的安检图像识别系统及其方法
CN112906660A (zh) * 2021-03-31 2021-06-04 浙江大华技术股份有限公司 安检预警方法和装置、存储介质及电子设备
CN114140542B (zh) * 2021-12-09 2022-11-22 成都智元汇信息技术股份有限公司 基于色彩补偿的切图方法、电子设备及介质
CN114140542A (zh) * 2021-12-09 2022-03-04 成都智元汇信息技术股份有限公司 基于色彩补偿的切图方法、电子设备及介质
CN115774031A (zh) * 2022-11-29 2023-03-10 同方威视科技江苏有限公司 查验任务处理方法、装置与电子设备
CN116244676A (zh) * 2023-05-09 2023-06-09 中航信移动科技有限公司 一种基于区块链的安全身份验证系统
CN116244676B (zh) * 2023-05-09 2023-07-07 中航信移动科技有限公司 一种基于区块链的安全身份验证系统
CN117011921A (zh) * 2023-08-18 2023-11-07 德科信息有限公司 一种提高人脸识别率的图像处理方法及系统
CN119049159A (zh) * 2024-09-10 2024-11-29 重庆赛力斯凤凰智创科技有限公司 目标车门的解锁方法、装置、电子设备

Also Published As

Publication number Publication date
CN109829362A (zh) 2019-05-31

Similar Documents

Publication Publication Date Title
WO2020124710A1 (zh) 安检辅助分析方法、装置、计算机设备和存储介质
CN111191532B (zh) 基于施工区域的人脸识别方法、装置、计算机设备
CN105518709B (zh) 用于识别人脸的方法、系统和计算机程序产品
CN111191539B (zh) 证件真伪验证方法、装置、计算机设备和存储介质
CN108229297B (zh) 人脸识别方法和装置、电子设备、计算机存储介质
Sanil et al. 2d-3d facial image analysis for identification of facial features using machine learning algorithms with hyper-parameter optimization for forensics applications
Gulhane et al. Advancing Facial Recognition: Enhanced Model with Improved Deepface Algorithm for Robust Adaptability in Diverse Scenarios
CN101558431A (zh) 脸认证设备
CN108241836A (zh) 用于安检的方法及装置
WO2018234384A1 (en) Detecting artificial facial images using facial landmarks
WO2018082011A1 (zh) 活体指纹识别方法及装置
Mortezaie et al. A survey on age-invariant face recognition methods
Ramkumar et al. An effectual facial expression recognition using HMM
EP3944188A1 (en) Image processing device, image processing method, and recording medium in which program is stored
Saib et al. Is face recognition with masks possible?
Meenakshi Real-Time Facial Recognition System—Design, Implementation and Validation
Nasrollahi et al. Haar-like features for robust real-time face recognition
Shivaleela Patil Composite sketch based face recognition using ANN classification
CN108875549A (zh) 图像识别方法、装置、系统及计算机存储介质
Rusia et al. A novel deep transfer learning-based approach for face pose estimation
Sarath Chandran et al. Facial emotion recognition system for unusual behaviour identification and alert generation
CN110674847A (zh) 卡片分类保存方法、装置、计算机设备和存储介质
HK40002811A (zh) 安检辅助分析方法、装置、计算机设备和存储介质
Veena et al. Cloud Computing Based Face Mask and Helmet Facial Detection for ATM Security Using Image Processing
Sannidhan et al. A comprehensive review on various state-of-the-art techniques for composite sketch matching

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 19899621

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205A DATED 30/09/2021)

122 Ep: pct application non-entry in european phase

Ref document number: 19899621

Country of ref document: EP

Kind code of ref document: A1