WO2019052318A1 - 一种电梯轿厢监控方法、装置及系统 - Google Patents

一种电梯轿厢监控方法、装置及系统 Download PDF

Info

Publication number
WO2019052318A1
WO2019052318A1 PCT/CN2018/101545 CN2018101545W WO2019052318A1 WO 2019052318 A1 WO2019052318 A1 WO 2019052318A1 CN 2018101545 W CN2018101545 W CN 2018101545W WO 2019052318 A1 WO2019052318 A1 WO 2019052318A1
Authority
WO
WIPO (PCT)
Prior art keywords
human body
point cloud
depth image
determining
elevator car
Prior art date
Application number
PCT/CN2018/101545
Other languages
English (en)
French (fr)
Inventor
龚晖
任烨
Original Assignee
杭州海康威视数字技术股份有限公司
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 杭州海康威视数字技术股份有限公司 filed Critical 杭州海康威视数字技术股份有限公司
Publication of WO2019052318A1 publication Critical patent/WO2019052318A1/zh

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
    • B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
    • B66B5/0006—Monitoring devices or performance analysers
    • B66B5/0012—Devices monitoring the users of the elevator system
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
    • B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B66—HOISTING; LIFTING; HAULING
    • B66B—ELEVATORS; ESCALATORS OR MOVING WALKWAYS
    • B66B5/00—Applications of checking, fault-correcting, or safety devices in elevators
    • B66B5/02—Applications of checking, fault-correcting, or safety devices in elevators responsive to abnormal operating conditions
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
    • G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
    • G06V40/103—Static body considered as a whole, e.g. static pedestrian or occupant recognition

Definitions

  • the present application relates to the field of computer vision technology, and in particular, to an elevator car monitoring method, device and system.
  • the elevator As an important vertical transportation tool, the elevator has been used more and more widely in life. Elevators are also convenient for people to travel, but also bring people a safety hazard. In one case, the above-mentioned safety hazards include the situation in which only children are boarding in the elevator car. Because the child is active and the safety precaution is weak, if it is not found in time, it is likely to cause a tragedy.
  • the image acquisition device is installed in the elevator car of many buildings, the video surveillance system in which the above image acquisition device is located can only provide the image acquisition function; subsequently, the monitoring personnel manually view the image collected by the image acquisition device. Identify if there is only a child in the elevator.
  • Such buildings may include, but are not limited to, communities, hotels, buildings, and the like.
  • the above-mentioned method of manually viewing images for monitoring requires the monitoring personnel to view the images collected by the image capturing device in real time, consumes manpower, is not intelligent enough, and may have omission phenomenon due to factors such as limited artificial energy or human negligence.
  • the purpose of the embodiment of the present application is to provide an elevator car monitoring method, device and system, so as to realize intelligent monitoring of the situation in which only children are used in the elevator, saving manpower, and avoiding possible artificial energy or human negligence. Factors, and omissions that occur.
  • the specific technical solutions are as follows:
  • an embodiment of the present application provides an elevator car monitoring method, the method comprising: obtaining a depth image acquired by an image acquisition device for an elevator car scene when detecting that the elevator car door is closed; determining the obtained Whether the human body target is included in the depth image; when determining that the human body target is included, determining the human body height information of the included human body target based on the obtained depth image; determining whether the determined human body height information is lower than a preset height threshold; When the determined human height information is lower than the preset height threshold, an alarm is issued.
  • the step of determining whether the obtained depth image includes a human body target comprises: extracting an image feature from the obtained depth image; determining, according to the extracted image feature and the first preset classifier, the obtained Whether the human body target is included in the depth image.
  • the step of determining whether the obtained depth image includes a human body target comprises: converting the obtained depth image into point cloud data in a world coordinate system; determining whether the point cloud data includes a human body target Corresponding point cloud data, wherein when it is determined that the point cloud data includes point cloud data corresponding to the human body target, the depth image obtained by the representation includes the human body target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to the human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The point cloud sub-data and the second preset classifier determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human body target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to the human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The point cloud data and the preset human body model determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the image acquisition device acquires a depth image by using the included depth image collection sub-device; and the step of converting the obtained depth image into point cloud data in a world coordinate system, comprising: obtaining the depth image Collecting parameter information of the sub-device, wherein the parameter information includes: focal length information, image main point information, installation height information, and installation angle information; using the focal length information, the image main point information, the obtained depth Converting the image to point cloud data in a device coordinate system, wherein the device coordinate system is: a coordinate system established based on an optical center of the depth image acquisition sub-device; using the installation height information and the installation angle information Converting point cloud data in the device coordinate system into point cloud data in the world coordinate system.
  • the parameter information includes: focal length information, image main point information, installation height information, and installation angle information
  • the device coordinate system is: a coordinate system established based on an optical center of the depth image acquisition sub-device; using the installation height information and the installation angle information Converting point cloud data in the device coordinate system into
  • the method further includes: determining whether the included human body target is one when determining to include the human body target; When the included human target is one, the step of determining the height information of the contained human target based on the obtained depth image is performed.
  • the method before the step of performing the alarm, the method further includes: determining, when determining that the determined human height information is lower than the preset height threshold, determining the number of included human targets; When the number of human targets is lower than the preset number, the step of performing the alarm is performed.
  • the method further comprises: controlling the elevator car to move to a designated floor; and controlling the elevator car after the elevator car moves to the designated floor Stopping the movement; controlling the elevator car to open the door after obtaining a door opening command for the elevator car.
  • an embodiment of the present application provides an elevator car monitoring device, where the device includes: a first obtaining module, configured to obtain an image capturing device for an elevator car scene when detecting that the elevator car door is closed a depth image obtained by the first determining module, configured to determine whether the obtained depth image includes a human body target, and a second determining module, configured to determine the included human body target based on the obtained depth image when determining to include the human body target
  • the human body height information is configured to determine whether the determined human body height information is lower than a preset height threshold
  • the alarm module is configured to: when determining that the determined human body height information is lower than the preset height threshold, Make an alarm.
  • the first determining module is specifically configured to extract an image feature from the obtained depth image; and determine, according to the extracted image feature and the first preset classifier, whether the obtained depth image includes a human body target .
  • the first determining module includes a converting unit and a determining unit, where the converting unit is configured to convert the obtained depth image into point cloud data in a world coordinate system, and the determining unit is configured to determine Whether the point cloud data corresponding to the human body target is included in the cloud data, wherein when the point cloud data includes the point cloud data corresponding to the human body target, the depth image obtained by the representation includes the human body target.
  • the determining unit is configured to perform clustering on the point cloud data to obtain various types of point cloud sub-data; and determine, according to the obtained types of point cloud sub-data and a second preset classifier, Whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the determining unit is specifically configured to cluster the point cloud data to obtain various types of point cloud sub-data; and determine the obtained based on the obtained types of point cloud sub-data and a preset human body model. Whether the point cloud sub-data corresponding to the human target is included in each type of point cloud sub-data.
  • the image capturing device collects the depth image by using the included depth image capturing sub-device;
  • the converting unit is specifically configured to obtain parameter information of the depth image acquiring sub-device, where the parameter information includes : focal length information, image main point information, mounting height information, and mounting angle information; converting the obtained depth image into point cloud data in a device coordinate system by using the focal length information and the image main point information, wherein
  • the device coordinate system is: a coordinate system established based on the optical center of the depth image acquisition sub-device; using the installation height information and the installation angle information, converting the point cloud data in the device coordinate system into Point cloud data in the world coordinate system.
  • the first obtaining module is specifically configured to obtain a depth image acquired by the image capturing device for an elevator car scene, and a color image corresponding to the obtained depth image; the first determining module, specifically And determining whether the obtained depth image includes a human body target by identifying whether the color image includes a human body target.
  • the device further includes a second determining module, where the second determining module is configured to: when determining the height information included in the human body target based on the obtained depth image, when determining to include the human body target, Determining whether the included human target is one; when determining that the included human target is one, triggering the second determining module.
  • the second determining module is configured to: when determining the height information included in the human body target based on the obtained depth image, when determining to include the human body target, Determining whether the included human target is one; when determining that the included human target is one, triggering the second determining module.
  • the device further includes a third determining module, configured to determine, when determining that the determined human height information is lower than the preset height threshold, before the performing the alarm The number of human targets is included; the alarm module is triggered when the determined number of contained human targets is less than a preset number.
  • a third determining module configured to determine, when determining that the determined human height information is lower than the preset height threshold, before the performing the alarm The number of human targets is included; the alarm module is triggered when the determined number of contained human targets is less than a preset number.
  • the device further includes a first control module, a second control module, and a third control module; the first control module, configured to control the elevator car to move to a designated floor after the alarm is performed
  • the second control module is configured to control the elevator car to stop moving after the elevator car moves to the designated floor;
  • the third control module is configured to obtain when the elevator car is obtained After the door opening command is issued, the elevator car is controlled to open the door.
  • an embodiment of the present application provides an electronic device, including a processor and a memory, where the memory is used to store a computer program, and the processor is configured to execute a computer program stored on the memory. Any of the above elevator car monitoring methods.
  • an embodiment of the present application provides an elevator system including: an elevator, and any of the above elevator car monitoring devices.
  • the elevator system includes a camera for collecting a depth image in the elevator car and transmitting to the elevator car monitoring device.
  • the embodiment of the present application provides a computer readable storage medium, where the computer readable storage medium stores a computer program, and when the computer program is executed by the processor, implements any of the elevator car monitoring described above. Method steps.
  • an embodiment of the present application provides a computer program product, when executed on a computer, causing a computer to perform the elevator car monitoring method steps of any of the above embodiments.
  • obtaining a depth image acquired by the image capturing device for the elevator car scene when detecting that the elevator car door is closed, obtaining a depth image acquired by the image capturing device for the elevator car scene; determining whether the obtained depth image includes a human body target; when determining that the human body target is included, Determining, according to the obtained depth image, human body height information of the included human body target; determining whether the determined human body height information is lower than a preset height threshold; and when determining that the determined human body height information is lower than a preset height threshold, performing an alarm .
  • the monitoring personnel can handle the corresponding treatment to avoid the danger of the child riding the elevator, so as to achieve only the child in the elevator.
  • the intelligent monitoring of the situation saves manpower.
  • the monitoring personnel are prevented from performing manual monitoring, there may be situations in which monitoring is omitted due to factors such as limited artificial energy or human negligence, thereby causing a tragedy.
  • implementing any of the products or methods of the present application necessarily does not necessarily require all of the advantages described above to be achieved at the same time.
  • FIG. 1 is a schematic flow chart of an elevator car monitoring method according to an embodiment of the present application.
  • FIG. 2 is another schematic flowchart of an elevator car monitoring method according to an embodiment of the present application.
  • FIG. 3 is a schematic structural diagram of an elevator car monitoring device according to an embodiment of the present application.
  • FIG. 4 is another schematic structural diagram of an elevator car monitoring device according to an embodiment of the present application.
  • FIG. 5 is a schematic structural diagram of an electronic device according to an embodiment of the present application.
  • the embodiment of the present application provides a method, a device and a system for monitoring an elevator car, so as to realize intelligent monitoring of a situation in which only a child is boarding in an elevator, saving manpower, and avoiding factors such as limited artificial energy or human negligence. And the omissions that have occurred.
  • an embodiment of the present application provides an elevator car monitoring method, which may include the following steps:
  • S101 Obtain a depth image acquired by the image capturing device for the elevator car scene when the elevator car door is detected to be closed.
  • the elevator car monitoring method provided by the embodiment of the present application can be applied to any electronic device that can obtain the depth image collected by the image capturing device, and the electronic device can be a computer or a smart phone. Monitor the server and more.
  • the image capturing device described above may be installed in the elevator car described above.
  • the image capturing device When the image capturing device is installed in the elevator car, it may be installed vertically or obliquely.
  • the vertical installation may be: the direction in which the elevator car is lifted and lowered, and the image capturing device may be installed on the top of the elevator car. In the middle, and facing the elevator car floor, the image is collected in the scene in the elevator car; the inclined installation may be installed in a direction that is at an angle with the direction in which the elevator car is moving up and down, and the image capturing device may be installed on At any corner of the top of the elevator car, image acquisition is performed on the scene in the elevator car. It can be understood that no matter how the image capturing device is installed, it is necessary to ensure that the image capturing device can capture images of all corners of the elevator car to the greatest extent.
  • the image capturing device may be any device that can collect depth information, that is, three-dimensional information, of the target in the scene, and may include, but is not limited to, a binocular depth camera, a TOF (Time of Flight) camera, and a structured optical camera.
  • depth information that is, three-dimensional information, of the target in the scene
  • the depth image may include depth information of each point in the elevator car scene, that is, distance information of each point in the elevator car scene to the image capturing device.
  • the pixel value of each pixel in the depth image is: depth information of a point in an elevator car scene corresponding to each pixel point. It can be seen that the depth image has characteristics that are not affected by factors such as illumination, shading and chromaticity.
  • the electronic device may first detect whether there is a target in the depth image, and the target may include a human body target and an object target. When it is detected that the target exists in the depth image, the electronic device determines each target included from the depth image, and subsequently identifies each target to determine whether each target is a human target.
  • the electronic device may pre-store the background depth image of the elevator car scene, that is, the depth image of the elevator car scene when the elevator car is idling; when the electronic device detects that the elevator car door is closed, the electronic device obtains The image acquisition device performs a difference image on the depth image acquired by the elevator car scene, and obtains a difference map. When there is a foreground pixel in the difference map, the image may be determined. The obtained depth image contains the target. Of course, when the obtained depth image is made to be different from the background depth image, it is necessary to ensure that the angle of view of the image capturing device does not change, that is, the position and/or angle of the image capturing device does not change.
  • the electronic device identifies each target, and the process of determining whether each target is a human target may be: matching each target with a preset three-dimensional human body model, and when the matching is successful, It can be determined that the target is a human target, and when the match fails, it can be determined that the target is not a human target.
  • the electronic device identifies each target, and the process of determining whether each target is a human target may also be: classifying each target in the depth image by using a pre-trained classifier, Determine if each target is a human target.
  • the pre-trained classifier may be: a machine learning model obtained by training based on a training sample and a preset algorithm, wherein the training sample may include a sample depth image of a human target and a non-human target.
  • the foregoing preset algorithm may be: a neural network algorithm including a convolutional neural network related algorithm.
  • the machine learning model may be: a neural network model.
  • the foregoing preset algorithm may be: a random forest algorithm.
  • the machine learning model may be: a random forest classification model.
  • the foregoing preset algorithm may be: a support vector machine algorithm.
  • the machine learning model may be: a support vector machine classification model, and the like.
  • the electronic device may determine the human body height information of the included human body target by using the depth image.
  • the electronic device may convert the obtained depth image into point cloud data in a world coordinate system, and subsequently, determine the human body of the human body target by using point cloud data in a world coordinate system corresponding to the human body target. Height information.
  • the highest point cloud data in the point cloud data in the world coordinate system corresponding to the human body target ie, the point cloud data corresponding to the top position of the human body head
  • the lowest point cloud data ie, the human target sole
  • the electronic device may sort the point cloud data in the world coordinate system corresponding to each human target based on the height of each point in the point cloud data in the world coordinate system corresponding to each human target. Calculating an average value of heights of a predetermined number of points in the sorting order as a first average value, and calculating an average value of heights of a predetermined number of points in the sorting order, as the second average value, the first average value is The absolute value of the second average value is used as the body height information of the human body target.
  • S104 Determine whether the determined human height information is lower than a preset height threshold.
  • S105 Perform an alarm when it is determined that the determined human height information is lower than a preset height threshold.
  • the electronic device may compare the determined human body height information with a preset height threshold to determine whether the determined human body height information is lower than a preset height threshold.
  • the depth image may include one or more human targets.
  • the human height information of each human target may be compared with a preset height threshold to determine each human body. Whether the target height information of the target is lower than the preset height threshold.
  • an alarm is issued.
  • the alarm may be performed in the form of outputting prompt information, or may be performed in the form of a prompting sound, or may be performed by changing the brightness of the screen, etc., which may cause The manner of monitoring the attention of the personnel can be used as the alarm form in the embodiment of the present application.
  • the alarm when the electronic device determines that the human body target included in the depth image is multiple, the alarm may be performed only when the human body height information of the included human body target is lower than the preset height threshold.
  • the preset height threshold may be set according to the height of the group of children, and in one implementation, the group of children may include a group of people under 10 years old.
  • the electronic device may directly use the image feature in the depth image to determine whether the depth image includes the human body target, or may use the feature of the point cloud data corresponding to the depth image to determine the depth image. Whether it is a human target or not, this is all right.
  • the step of determining whether the obtained depth image includes a human body target may include: extracting an image feature from the obtained depth image; and based on the extracted image feature and the first pre- A classifier is provided to determine whether the obtained depth image contains a human body target.
  • the first preset classifier may be: a machine learning model obtained by training based on a training sample and a preset algorithm, where the training sample may be: a sample depth image including a human target and a non-human target, ie,
  • the image features utilized by the first predetermined classifier may be: image features in the sample depth image.
  • the foregoing preset algorithm may be: a neural network algorithm including a convolutional neural network related algorithm.
  • the machine learning model may be: a neural network model.
  • the foregoing preset algorithm may be: a random forest algorithm.
  • the machine learning model may be: a random forest classification model.
  • the foregoing preset algorithm may be: a support vector machine algorithm.
  • the machine learning model may be: a support vector machine classification model, and the like.
  • the electronic device may perform image feature extraction on the depth image, for example, performing edge feature extraction on the depth image by using a Canny operator or the like. Subsequently, the extracted image features are input into a first preset classifier, and the first preset classifier classifies the extracted image features to determine whether the human body target is included in the depth image.
  • human body image features may be pre-stored in the electronic device. After the electronic device extracts the image features from the depth image, the extracted image features are matched with the pre-stored human body image features. When the matching is successful, the depth image may be determined to include the human body target.
  • the step of determining whether the obtained depth image includes a human body target may include: converting the obtained depth image into point cloud data in a world coordinate system; determining a point cloud Whether the point cloud data corresponding to the human body target is included in the data, wherein when the point cloud data includes the point cloud data corresponding to the human body target, the depth image obtained by the representation includes the human body target.
  • the electronic device may first convert the depth image into point cloud data in the world coordinate system, and then determine whether the human body target is included in the depth image based on the point cloud data in the world coordinate system.
  • the image acquisition device may use the included depth image acquisition sub-device to collect the depth image.
  • the depth image collection device in the image acquisition device that collects the depth image may be utilized.
  • the parameter information of the device converts the depth image into point cloud data in the world coordinate system.
  • the parameter information may include an internal parameter and an external parameter of the depth image collection sub-device, and the electronic device may calibrate the image collection device according to the Zhang Zhengyou calibration method to determine the parameter information, or the electronic device may directly send the image collection device to the factory.
  • the calibrated internal reference is used as an internal parameter in the parameter information mentioned in the embodiment of the present application, and the external parameter can be obtained by measurement and pre-calibration, which is all possible.
  • the step of converting the obtained depth image into point cloud data in a world coordinate system may include: obtaining parameter information of the depth image collection sub-device, wherein the parameter information includes: focal length information , such as the main point information, the installation height information, and the installation angle information; using the focal length information, like the main point information, converting the obtained depth image into point cloud data in the device coordinate system, wherein the device coordinate system can be: based on the depth The coordinate system established by the optical center of the image acquisition sub-device; the point cloud data in the device coordinate system is converted into point cloud data in the world coordinate system by using the installation height information and the installation angle information.
  • the parameter information includes: focal length information , such as the main point information, the installation height information, and the installation angle information
  • the focal length information like the main point information
  • the image principal point can be: the intersection of the optical axis of the depth image acquisition sub-device and the image plane.
  • the image principal point information may include: two-dimensional coordinates of the principal point in the depth image. Wherein, using the two-dimensional coordinates of the principal point, the two-dimensional coordinates of each pixel in the depth image can be determined.
  • the two-dimensional coordinates (u, v) of each pixel in the depth image are converted into three-dimensional coordinates (X C , Y C , Z C ) in the device coordinate system to obtain a depth image.
  • Point cloud data The preset three-dimensional rectangular coordinate system may be: a coordinate system established based on the optical center of the depth image acquisition sub-device, and the formula used in the coordinate conversion is as follows:
  • f Dx and f Dy are the focal lengths of the depth image acquisition sub-device; (u D0 , v D0 ) is the two-dimensional coordinates of the image main point; Z C is the pixel point (u, v) corresponding to the depth image The point in the scene, the distance information to the depth image acquisition sub-device, that is, the pixel value of the pixel point (u, v).
  • f Dx represents the focal length in the x-axis direction of the determined depth image acquisition sub-device
  • f Dy represents the determined focal length in the y-axis direction of the depth image acquisition sub-device.
  • the above f Dx and f Dy are all included in the above focal length information, and can be directly determined by Zhang Zhengyou calibration method.
  • the above two-dimensional coordinates of the main point can also be directly determined by Zhang Zhengyou calibration method.
  • the above device coordinate system may be a three-dimensional Cartesian coordinate system.
  • the installation height information and the installation angle information of the depth image acquisition sub-device including the elevation angle and the deflection angle of the depth image acquisition sub-device, determine the conversion relationship between the device coordinate system and the world coordinate system, and then based on the determined conversion The relationship converts point cloud data in the device coordinate system into point cloud data in the world coordinate system.
  • the point cloud data may be clustered by using a preset clustering algorithm to determine each target included in the depth image, and further And determining whether each target is a human target, in a case, the step of determining whether the point cloud data includes the point cloud data corresponding to the human target may include: clustering the point cloud data to obtain various point clouds And determining, according to the obtained types of point cloud sub-data and the second preset classifier, whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human body target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to the human body target may include: clustering the point cloud data to obtain various point cloud sub-data; The point cloud sub-data and the preset human body model determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human body target.
  • the foregoing predetermined clustering algorithm may be a density-based clustering algorithm, for example, a DBSCAN (Density-Based Spatial Clustering of Applications with Noise) clustering algorithm.
  • the DBSCAN clustering algorithm requires two parameters, namely the scan radius (eps) and the minimum inclusion point (minPts).
  • the electronic device uses the DBSCAN clustering algorithm to select an unvisited point from the point cloud data (ie, is not marked as an accessed point or is not marked as noise).
  • the electronic device determines whether the number of points near the current point is greater than or equal to minPts, if The electronic device determines that the number of nearby points of the current point is greater than or equal to minPts, the electronic device determines the current point and the determined nearby point as a cluster, and the electronic device marks the current point as visited; the electronic device targets the The cluster is recursively processed in the manner described above for all points in the cluster that are not marked as visited or not marked as noise, thereby expanding the cluster; if the cluster is fully expanded, ie all points within the cluster are Marked as accessed; the electronic device re-selects an unvisited point as the current point, performing the subsequent process.
  • the electronic device determines that the number of points near the current point is less than minPts, the electronic device marks the current point as noise; until the point cloud data does not include the unvisited point, the clustering is completed. At this time, the electronic device can determine various types of point cloud sub-data included in the point cloud data. Subsequently, the electronic device can process each type of point cloud sub-data to determine whether each type of point cloud sub-data is a human body. Point cloud subdata corresponding to the target.
  • the electronic device may input each type of point cloud sub-data obtained into the second preset classifier, so that the second preset classifier classifies each type of point cloud sub-data, and determines Whether each type of point cloud subdata is a point cloud subdata corresponding to a human target.
  • the second preset classifier may be: a machine learning model obtained by training based on a training sample and a preset algorithm, where the training sample may be: a sample depth image including a human target and a non-human target. Point cloud data.
  • the foregoing preset algorithm may be: a neural network algorithm including a convolutional neural network related algorithm.
  • the machine learning model may be: a neural network model.
  • the foregoing preset algorithm may be: a random forest algorithm.
  • the machine learning model may be: a random forest classification model.
  • the foregoing preset algorithm may be: a support vector machine algorithm.
  • the machine learning model may be: a support vector machine classification model, and the like.
  • the electronic device may match each type of point cloud data obtained by the preset human body model, and when the matching is successful, determine the point cloud sub data as the point cloud sub data corresponding to the human target. When the matching fails, it is determined that the point cloud sub data is not the point cloud sub data corresponding to the human target.
  • the foregoing preset clustering algorithm may also be a grid-based clustering algorithm, such as a STING clustering algorithm, or a partitioning-based clustering algorithm, for example, K-means clustering.
  • the algorithm and the like are not limited to the algorithm that can cluster the point cloud data. Any algorithm that can cluster the point cloud data can be applied to the embodiment of the present application.
  • whether a human body target is included in the depth image may be determined by using a color image corresponding to the depth image acquired by the elevator car, wherein the obtaining the depth image acquired by the image capturing device for the elevator car scene
  • the step of obtaining the depth image acquired by the image capturing device for the elevator car scene and the color image corresponding to the obtained depth image; and the step of determining whether the obtained depth image includes the human body target may include : determining whether the obtained depth image contains a human body target by identifying whether the color image includes a human body target.
  • the face recognition algorithm may be used to identify whether the human body target is included in the color image to determine whether the depth image includes a human body target, wherein when the color image includes the human body target, the depth image includes the human body target; when the color image does not When a human target is included, the depth target does not contain a human target.
  • the color image may be an image of any color mode, for example, an image of an RGB (Red Green Blue) color mode, or a YUV color mode image, and the like.
  • RGB Red Green Blue
  • YUV also known as YCrCb
  • Y means brightness (Luminance or Luma), that is, gray level value
  • U and "V” means color (Chrominance or Chroma)
  • Chroma Chrominance or Chroma
  • the method may include the following steps:
  • S201 Obtain a depth image acquired by the image acquisition device for the elevator car scene when the elevator car door is detected to be closed.
  • the above S201 is the same as S101 shown in FIG. 1, and the above S202 is the same as S102 shown in FIG.
  • S205 Determine whether the determined human height information is lower than a preset height threshold.
  • the above S204 is the same as S103 shown in FIG. 1
  • the above S205 is the same as S104 shown in FIG. 1
  • the above S206 is the same as S105 shown in FIG.
  • the electronic device determines that the human body height information of the human body target included in the elevator car is lower than a preset height threshold, that is, when determining that the human body target included in the elevator car is a child, and when When the number of human targets is lower than the preset number, it is necessary to remind the monitoring personnel to pay attention, and the monitoring personnel take corresponding measures against the above situations to avoid the above-mentioned human target, that is, the child is in danger.
  • the method may further include: determining the number of the included human target when determining that the determined human height information is lower than a preset height threshold; and determining the included human target The step of performing the alarm is performed when the number of the number is lower than the preset number.
  • the step of determining the number of human targets included in the foregoing may be performed between the step of determining whether the determined human height information is lower than a preset height threshold and the step of performing an alarm, that is, determining the determined height information of the human body.
  • the preset height threshold is lower, the number of included human targets is determined; when the determined number of included human targets is lower than the preset number, an alarm is issued.
  • the preset number may be any positive integer.
  • the preset number may be 3, that is, when the electronic device determines that the number of human targets included in the elevator car is less than 3, that is, the number of children is When two or one, the electronic device gives an alarm.
  • the preset number may be 2, that is, when the electronic device determines that the number of human targets included in the elevator car is less than 2, that is, the number of children is one, the electronic device performs an alarm.
  • the method may further include: controlling the elevator car to move to the designated floor; and controlling the elevator car after the elevator car moves to the designated floor The car stops moving; when the door opening command for the elevator car is obtained, the elevator car is controlled to open the door.
  • the electronic device after the electronic device performs an alarm, it can receive the movement control command of the monitoring personnel for the elevator car, or the electronic device performs an alarm, automatically triggers the movement control instruction, and the like, which is all possible. After the electronic device obtains the above-mentioned movement control command, the electronic device can control the elevator car to move to the designated floor and control the elevator car to stop at the designated layer.
  • the designated layer may be a pre-set one floor, such as a floor; or may be a floor in the direction of the elevator car moving and closest to the current position of the elevator car, where the elevator car is currently located
  • the location may be: the location of the elevator car when the electronic device obtains the above mobile control command.
  • the moving direction of the elevator car may include rising or falling.
  • the car door of the elevator car can be controlled to be in a closed state, and after the door opening command for the elevator car is obtained, the elevator car is controlled to open the door.
  • the embodiment of the present application further provides an elevator car monitoring device.
  • the device may include: a first obtaining module 310, configured to detect when the elevator car door is closed. Obtaining a depth image acquired by the image acquisition device for the elevator car scene; the first determining module 320 is configured to determine whether the obtained depth image includes a human body target; and the second determining module 330 is configured to determine that the human body target is included And determining, according to the obtained depth image, human body height information of the included human body target; the first determining module 340 is configured to determine whether the determined human body height information is lower than a preset height threshold; and the alarm module 350 is configured to determine When the determined human height information is lower than the preset height threshold, an alarm is issued.
  • the first determining module 320 is specifically configured to extract an image feature from the obtained depth image; and determine, in the obtained depth image, based on the extracted image feature and the first preset classifier. Whether it contains human targets.
  • the first determining module 320 includes a converting unit and a determining unit, and the converting unit is configured to convert the obtained depth image into point cloud data in a world coordinate system; the determining unit, And determining, by the point cloud data, whether the point cloud data corresponding to the human body target is included, wherein when the point cloud data corresponding to the point cloud data corresponding to the human body target is included, the depth image obtained by the representation includes the human body target.
  • the determining unit is specifically configured to cluster the point cloud data to obtain various types of point cloud sub-data; and based on the obtained types of point cloud sub-data and a second preset classifier And determining whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the determining unit is specifically configured to cluster the point cloud data to obtain various types of point cloud sub-data; and determine, according to the obtained types of point cloud sub-data and a preset human body model, Whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the image acquisition device uses the included depth image collection sub-device to acquire a depth image; the conversion unit is specifically configured to obtain parameter information of the depth image collection sub-device, wherein the parameter The information includes: focal length information, image main point information, installation height information, and installation angle information; and the obtained depth image is converted into point cloud data in a device coordinate system by using the focal length information and the image principal point information,
  • the device coordinate system is: a coordinate system established based on an optical center of the depth image acquisition sub-device; using the installation height information and the installation angle information, point cloud data in the device coordinate system , converted to point cloud data in the world coordinate system.
  • the first obtaining module 310 is specifically configured to obtain a depth image acquired by the image capturing device for an elevator car scene, and a color image corresponding to the obtained depth image;
  • the determining module 320 is specifically configured to determine whether the obtained depth image includes a human body target by identifying whether the color image includes a human body target.
  • the device may further include a second determining module 410, where the second determining module 410 is configured to determine the included human target based on the obtained depth image. Before the height information, when it is determined that the human body target is included, it is determined whether the included human body target is one; when it is determined that the included human body target is one, the second determining module 330 is triggered.
  • the device may further include a third determining module, configured to determine, when the alarm is performed, that the determined human height information is lower than the preset height threshold The number of contained human targets is determined; the alarm module 350 is triggered when the determined number of contained human targets is less than a preset number.
  • a third determining module configured to determine, when the alarm is performed, that the determined human height information is lower than the preset height threshold The number of contained human targets is determined; the alarm module 350 is triggered when the determined number of contained human targets is less than a preset number.
  • the device may further include a first control module, a second control module, and a third control module, where the first control module is configured to control the elevator car after the alarm is performed Moving to a designated floor; the second control module is configured to control the elevator car to stop moving after the elevator car moves to the designated floor; the third control module is configured to obtain a target After the door opening command of the elevator car is described, the elevator car is controlled to open the door.
  • the embodiment of the present application further provides an elevator system, including: an elevator, and any of the above elevator car monitoring devices.
  • the elevator system includes a camera for collecting a depth image in the elevator car and transmitting to the elevator car monitoring device.
  • the embodiment of the present application further provides an electronic device, including a processor and a memory, wherein the memory is used to store a computer program, and the processor is configured to execute a computer stored on the memory.
  • the memory is used to store a computer program
  • the processor is configured to execute a computer stored on the memory.
  • any of the above elevator car monitoring methods are implemented.
  • an embodiment of the present invention further provides an electronic device, as shown in FIG. 5, including a processor 510, a communication interface 520, a memory 530, and a communication bus 540, wherein the processor 510 and the communication interface 520
  • the memory 530 is configured to communicate with each other through the communication bus 540.
  • the memory 530 is configured to store a computer program.
  • the processor 510 is configured to execute any of the programs provided in the embodiments of the present application when the computer program stored in the memory 530 is executed.
  • the elevator car monitoring method may include the following steps: when detecting that the elevator car door is closed, obtaining a depth image acquired by the image capturing device for the elevator car scene; determining whether the obtained depth image includes the human body a target; when determining that the human body target is included, determining human body height information of the included human body target based on the obtained depth image; determining whether the determined human body height information is lower than a preset height threshold; and determining that the determined human body height information is low At the preset height threshold, an alarm is issued.
  • the step of determining whether the obtained depth image includes a human body target comprises: extracting an image feature from the obtained depth image; and based on the extracted image feature and the first preset classifier, It is determined whether the obtained depth image contains a human body target.
  • the step of determining whether the obtained depth image includes a human body target comprises: converting the obtained depth image into point cloud data in a world coordinate system; determining whether the point cloud data is included The point cloud data corresponding to the human body target is included, wherein when the point cloud data corresponding to the point cloud data corresponding to the human body target is included, the depth image obtained by the representation includes the human body target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to a human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The obtained point cloud sub-data and the second preset classifier determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to a human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The obtained point cloud sub-data and the preset human body model determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the image capturing device collects a depth image by using the included depth image capturing sub-device; and the step of converting the obtained depth image into point cloud data in a world coordinate system, including: obtaining The parameter information of the depth image collection sub-device, wherein the parameter information includes: focal length information, image main point information, installation height information, and installation angle information; using the focal length information, the image main point information, The obtained depth image is converted into point cloud data in a device coordinate system, wherein the device coordinate system is: a coordinate system established based on an optical center of the depth image acquisition sub-device; using the installation height information and the The angle information is installed, and the point cloud data in the device coordinate system is converted into point cloud data in the world coordinate system.
  • the step of obtaining a depth image acquired by the image acquisition device for the elevator car scene includes: obtaining a depth image acquired by the image acquisition device for the elevator car scene, and the obtained depth a color image corresponding to the image; the step of determining whether the obtained depth image includes a human body target comprises: determining whether the obtained depth image includes a human body target by identifying whether the color image includes a human body target.
  • the processor 510 is further configured to: when the step of determining the height information of the included human body target based on the obtained depth image, determining that the human body target is included Whether the human target is one or not; when it is judged that the included human target is one, performing the step of determining the height information of the included human target based on the obtained depth image.
  • the processor 510 is further configured to: determine, before the step of performing the alarm, determining that the included human body is determined when the determined human body height information is lower than the preset height threshold The number of targets; when the determined number of included human targets is lower than a preset number, the step of performing the alarm is performed.
  • the processor 510 is further configured to: after the step of performing an alarm, controlling the elevator car to move to a designated floor; when the elevator car moves to the designation After the floor, the elevator car is controlled to stop moving; after the door opening command for the elevator car is obtained, the elevator car is controlled to open the door.
  • the communication bus mentioned in the above electronic device may be a Peripheral Component Interconnect (PCI) bus or an Extended Industry Standard Architecture (EISA) bus.
  • PCI Peripheral Component Interconnect
  • EISA Extended Industry Standard Architecture
  • the communication bus can be divided into an address bus, a data bus, a control bus, and the like. For ease of representation, only one thick line is shown in the figure, but it does not mean that there is only one bus or one type of bus.
  • the communication interface is used for communication between the above electronic device and other devices.
  • the memory may include a random access memory (RAM), and may also include a non-volatile memory (NVM), such as at least one disk storage.
  • RAM random access memory
  • NVM non-volatile memory
  • the memory may also be at least one storage device located away from the aforementioned processor.
  • the above processor may be a general-purpose processor, including a central processing unit (CPU), a network processor (NP), etc.; or may be a digital signal processing (DSP), dedicated integration.
  • CPU central processing unit
  • NP network processor
  • DSP digital signal processing
  • ASIC Application Specific Integrated Circuit
  • FPGA Field-Programmable Gate Array
  • the embodiment of the present application provides a computer readable storage medium, where the computer readable storage medium stores a computer program, and when the computer program is executed by the processor, the embodiment of the present application is provided.
  • the method may include the steps of: obtaining a depth image acquired by the image capturing device for the elevator car scene when detecting that the elevator car door is closed; determining the obtained depth image Whether the human body target is included; when it is determined that the human body target is included, determining the human body height information of the contained human body target based on the obtained depth image; determining whether the determined human body height information is lower than a preset height threshold; When the human body height information is lower than the preset height threshold, an alarm is issued.
  • the step of determining whether the obtained depth image includes a human body target comprises: extracting an image feature from the obtained depth image; and based on the extracted image feature and the first preset classifier, It is determined whether the obtained depth image contains a human body target.
  • the step of determining whether the obtained depth image includes a human body target comprises: converting the obtained depth image into point cloud data in a world coordinate system; determining whether the point cloud data is included The point cloud data corresponding to the human body target is included, wherein when the point cloud data corresponding to the point cloud data corresponding to the human body target is included, the depth image obtained by the representation includes the human body target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to a human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The obtained point cloud sub-data and the second preset classifier determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the step of determining whether the point cloud data includes point cloud data corresponding to a human body target comprises: clustering the point cloud data to obtain various point cloud sub-data; The obtained point cloud sub-data and the preset human body model determine whether the obtained point cloud sub-data includes the point cloud sub-data corresponding to the human target.
  • the image capturing device collects a depth image by using the included depth image capturing sub-device; and the step of converting the obtained depth image into point cloud data in a world coordinate system, including: obtaining The parameter information of the depth image collection sub-device, wherein the parameter information includes: focal length information, image main point information, installation height information, and installation angle information; using the focal length information, the image main point information, The obtained depth image is converted into point cloud data in a device coordinate system, wherein the device coordinate system is: a coordinate system established based on an optical center of the depth image acquisition sub-device; using the installation height information and the The angle information is installed, and the point cloud data in the device coordinate system is converted into point cloud data in the world coordinate system.
  • the step of obtaining a depth image acquired by the image acquisition device for the elevator car scene includes: obtaining a depth image acquired by the image acquisition device for the elevator car scene, and the obtained depth a color image corresponding to the image; the step of determining whether the obtained depth image includes a human body target comprises: determining whether the obtained depth image includes a human body target by identifying whether the color image includes a human body target.
  • the following steps may be further implemented: before the step of determining the height information of the included human body target based on the obtained depth image, when determining to include the human body target And determining whether the included human target is one; and when determining that the included human target is one, performing the step of determining the height information of the included human target based on the obtained depth image.
  • the following steps may be further implemented: before the step of performing the alarm, determining that the determined human height information is lower than the preset height threshold, determining The number of human targets included; when the determined number of included human targets is lower than a preset number, the step of performing an alarm is performed.
  • the following steps may be further implemented: after the step of performing an alarm, controlling the elevator car to move to a designated floor; when the elevator car moves to After the designated floor, the elevator car is controlled to stop moving; after the door opening instruction for the elevator car is obtained, the elevator car is controlled to open the door.
  • the embodiment of the present application provides a computer program product, when it is run on a computer, causing the computer to execute the elevator car monitoring method according to any one of the above embodiments.
  • the solution provided by the embodiment of the present application when detecting that the elevator car door is closed, obtaining a depth image acquired by the image capturing device for the elevator car scene, and determining whether the depth image includes a human body target, and when determining that the human body target is included, Further determining the human body height information of the human body target, and combining the preset height threshold value to determine whether the included human body target is a child, and determining the included human body target when determining that the human body height information of the human body target is lower than a preset height threshold value For the child, it can be determined that there is only a child riding in the elevator car at this time, that is, only the child is riding the elevator.

Landscapes

  • Engineering & Computer Science (AREA)
  • Human Computer Interaction (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Multimedia (AREA)
  • Theoretical Computer Science (AREA)
  • Image Processing (AREA)
  • Image Analysis (AREA)
  • Indicating And Signalling Devices For Elevators (AREA)

Abstract

一种电梯轿厢监控方法、装置及系统，该方法包括：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像（S101）；判断所获得的深度图像中是否包含人体目标（S102）；当判断包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息（S103）；判断所确定的人体高度信息是否低于预设高度阈值（S104）；当判断所确定的人体高度信息低于预设高度阈值时，进行报警（S105）。实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力，并避免了可能存在因为人工精力有限或者人为疏忽等因素，而出现的遗漏现象。

Description

一种电梯轿厢监控方法、装置及系统
本申请要求于2017年9月13日提交中国专利局、申请号为201710822052.0发明名称为“一种电梯轿厢监控方法、装置及系统”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
本申请涉及计算机视觉技术领域，特别是涉及一种电梯轿厢监控方法、装置及系统。
背景技术
电梯作为重要的垂直运输工具，在生活中得到了越来越广泛的应用。电梯在方便人们出行的同时，也给人们带来了安全隐患。在一种情况中，上述安全隐患包括电梯轿厢内仅有小孩搭乘的情况，由于小孩好动且安全防范意识薄弱，如不能及时发现，很可能会酿成惨剧。虽然在很多建筑的电梯轿厢内都安装了图像采集设备，然而目前上述图像采集设备所在的视频监控系统仅能提供图像采集功能；后续的，监控人员通过人工观看图像采集设备所采集的图像，辨识是否出现电梯内仅有小孩搭乘的情况。上述建筑可以包括但不限于小区、宾馆和大厦等。
采用上述人工观看图像进行监控的方式，需要监控人员实时地观看图像采集设备所采集的图像，消耗人力，不够智能，而且可能存在因为人工精力有限或者人为疏忽等因素，而出现的遗漏现象。
发明内容
本申请实施例的目的在于提供一种电梯轿厢监控方法、装置及系统，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力，并避免可能存在因为人工精力有限或者人为疏忽等因素，而出现的遗漏现象。具体技术方案如下：
一方面，本申请实施例提供了一种电梯轿厢监控方法，所述方法包括：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；确定所获得的深度图像中是否包含人体目标；当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；判断所确定的人体高度信息是否低于预设高度阈值；当判断所确定的人体高度信 息低于所述预设高度阈值时，进行报警。
可选地，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
可选地，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：将所获得的深度图像转换为世界坐标系下的点云数据；确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
可选地，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
可选地，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
可选地，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，包括：获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
可选地，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，包括：获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；所述确定所获得的深度图像中是否包含人体目标的步骤，包括：通过识别所述色彩图像中是否包含人体目标， 确定获得的深度图像中是否包含人体目标。
可选地，在所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤之前，所述方法还包括：当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，执行所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤。
可选地，在所述进行报警的步骤之前，所述方法还包括：当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
可选地，在所述进行报警的步骤之后，所述方法还包括：控制所述电梯轿厢移动至指定楼层；当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
另一方面，本申请实施例提供了一种电梯轿厢监控装置，所述装置包括：第一获得模块，用于当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；第一确定模块，用于确定所获得的深度图像中是否包含人体目标；第二确定模块，用于当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；第一判断模块，用于判断所确定的人体高度信息是否低于预设高度阈值；报警模块，用于当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
可选地，所述第一确定模块，具体用于从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
可选地，所述第一确定模块包括转换单元和确定单元；所述转换单元，用于将所获得的深度图像转换为世界坐标系下的点云数据；所述确定单元，用于确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
可选地，所述确定单元，具体用于对所述点云数据进行聚类，获得各类 点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
可选地，所述确定单元，具体用于对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
可选地，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；所述转换单元，具体用于获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
可选地，所述第一获得模块，具体用于获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；所述第一确定模块，具体用于通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
可选地，所述装置还包括第二判断模块；所述第二判断模块，用于在所述基于所获得的深度图像，确定所包含人体目标的高度信息之前，当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，触发所述第二确定模块。
可选地，所述装置还包括第三确定模块；所述第三确定模块，用于在所述进行报警之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，触发所述报警模块。
可选地，所述装置还包括第一控制模块、第二控制模块和第三控制模块；所述第一控制模块，用于在所述进行报警之后，控制所述电梯轿厢移动至指定楼层；所述第二控制模块，用于当所述电梯轿厢移动至所述指定楼层后， 控制所述电梯轿厢停止移动；所述第三控制模块，用于当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
另一方面，本申请实施例提供一种电子设备，包括处理器和存储器，其中，所述存储器，用于存放计算机程序；所述处理器，用于执行存储器上所存放的计算机程序时，实现上述任一电梯轿厢监控方法。
另一方面，本申请实施例提供一种电梯系统，包括：电梯，以及上述任一电梯轿厢监控装置。
可选地，该电梯系统，包括摄像头，用于采集电梯轿厢内的深度图像，并发送给所述电梯轿厢监控装置。
另一方面，本申请实施例提供一种计算机可读存储介质，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现上述任一所述的电梯轿厢监控方法步骤。
另一方面，本申请实施例提供一种计算机程序产品，当其在计算机上运行时，使得计算机执行上述实施例中任一所述的电梯轿厢监控方法步骤。
本申请实施例中，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；判断所获得的深度图像中是否包含人体目标；当判断包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；判断所确定的人体高度信息是否低于预设高度阈值；当判断所确定的人体高度信息低于预设高度阈值时，进行报警。
当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了监控人员进行人工监控时，可 能存在因为人工精力有限或者人为疏忽等因素，而出现监控遗漏的情况，进而造成出现惨剧的情况。当然，实施本申请的任一产品或方法必不一定需要同时达到以上所述的所有优点。
附图说明
为了更清楚地说明本申请实施例或现有技术中的技术方案，下面将对实施例或现有技术描述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他的附图。
图1为本申请实施例所提供的一种电梯轿厢监控方法的流程示意图。
图2为本申请实施例所提供的一种电梯轿厢监控方法的另一流程示意图。
图3为本申请实施例所提供的一种电梯轿厢监控装置的结构示意图。
图4为本申请实施例所提供的一种电梯轿厢监控装置的另一结构示意图。
图5为本申请实施例所提供的一种电子设备的结构示意图。
具体实施方式
下面将结合本申请实施例中的附图，对本申请实施例中的技术方案进行清楚、完整地描述，显然，所描述的实施例仅仅是本申请一部分实施例，而不是全部的实施例。基于本申请中的实施例，本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例，都属于本申请保护的范围。
本申请实施例提供了一种电梯轿厢监控方法、装置及系统，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力，并避免可能存在因为人工精力有限或者人为疏忽等因素，而出现的遗漏现象。
如图1所示，本申请实施例提供了一种电梯轿厢监控方法，可以包括如下步骤：
S101：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像。
可以理解的是，本申请实施例所提供的一种电梯轿厢监控方法，可以应用于任一可以获得图像采集设备所采集的深度图像的电子设备中，上述电子 设备可以是电脑、智能手机、监控服务器等等。
上述图像采集设备可以安装于上述电梯轿厢内。在电梯轿厢内安装上述图像采集设备时，可以垂直安装，也可以倾斜安装，其中，垂直安装可以是：延电梯轿厢升降的方向进行安装，上述图像采集设备可以安装于电梯轿厢顶部的中间，且正对电梯轿厢地面，以对电梯轿厢内的场景进行图像采集；倾斜安装可以是以与电梯轿厢升降的方向存在一定夹角的方向进行安装，上述图像采集设备可以安装于电梯轿厢顶部的任一边角处，以对电梯轿厢内的场景进行图像采集。可以理解的是，不管如何安装上述图像采集设备，需保证上述图像采集设备，可以最大程度的采集到电梯轿厢内各个角落的图像。
上述图像采集设备可以为任一可以采集场景中目标的深度信息也即三维信息的设备，可以包括但不限于双目深度摄像机、TOF(Time of Flight，飞行时间)摄像机及结构光摄像机。
上述深度图像中可以包含电梯轿厢场景中各点的深度信息，即电梯轿厢场景中各点到图像采集设备的距离信息。在一种实现方式中，上述深度图像中各像素点的像素值即为：各像素点对应的电梯轿厢场景中的点的深度信息。可见，深度图像具有不受光照、阴影及色度等因素影响的特性。
S102：确定所获得的深度图像中是否包含人体目标。
电子设备获得上述深度图像后，可以从上述深度图像中识别是否包含人体目标。在一种实现方式中，电子设备可以首先检测上述深度图像中是否存在目标，上述目标可以包括人体目标和物体目标。当检测到上述深度图像中存在目标时，电子设备从上述深度图像中确定出所包含的每一目标，后续的，针对每一目标进行识别，判断每一目标是否为人体目标。
一种情况中，电子设备可以预存有上述电梯轿厢场景的背景深度图像，即为电梯轿厢为空载时，电梯轿厢场景的深度图像；电子设备检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，将所获得的深度图像与上述背景深度图像进行作差，获得差值图，当差值图中存在前景像素点时，则可以确定上述所获得的深度图像中包含目标。当然，在将所获得的深度图像与上述背景深度图像进行作差时，需保证图像采集设备 的视角未发生变化，即图像采集设备的位置和/或角度未发生变化。
在一种实现方式中，电子设备针对每一目标进行识别，判断每一目标是否为人体目标的过程可以是：将每一目标与预先设置的三维立体人体模型进行匹配，当匹配成功时，则可以确定目标为人体目标，当匹配失败时，则可以确定目标不为人体目标。
在另一种实现方式中，电子设备针对每一目标进行识别，判断每一目标是否为人体目标的过程也可以是：利用预先训练的分类器，对上述深度图像中的每一目标进行分类，确定每一目标是否为人体目标。可以理解的是，上述预先训练的分类器可以是：基于训练样本以及预设算法训练所得的机器学习模型，其中，上述训练样本可以包含人体目标和非人体目标的样本深度图像。对于上述预设算法来说，上述预设算法可以为：包括卷积神经网络相关算法在内的神经网络算法，此时，上述机器学习模型可以为：神经网络模型。或者，上述预设算法可以为：随机森林算法，此时，上述机器学习模型可以为：随机森林分类模型。或者，上述预设算法可以为：支持向量机算法，此时，上述机器学习模型可以为：支持向量机分类模型，等等。
S103：当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息。
当电子设备确定深度图像中包含人体目标时，电子设备可以利用上述深度图像，确定所包含人体目标的人体高度信息。在一种实现方式中，电子设备可以将所获得的深度图像转化为世界坐标系下的点云数据，后续的，利用人体目标所对应世界坐标系下的点云数据，确定该人体目标的人体高度信息。
在一种情况中，可以是确定人体目标所对应世界坐标系下的点云数据中最高的点云数据(即人体目标头顶部位对应的点云数据)及最低的点云数据(即人体目标脚底部位对应的点云数据)，将上述最高的点云数据与最低的点云数据之间的高度差，作为该人体目标的人体高度信息。
在另一种情况中，电子设备可以针对每一人体目标所对应世界坐标系下的点云数据，基于每一人体目标所对应世界坐标系下的点云数据中的每一点的高度进行排序，计算排序顺序中前预定数量个点的高度的平均值，作为第 一平均值，并计算排序顺序中后预定数量个点的高度的平均值，作为第二平均值，将上述第一平均值与第二平均值的绝对值作为该人体目标的人体高度信息。
S104：判断所确定的人体高度信息是否低于预设高度阈值。
S105：当判断所确定的人体高度信息低于预设高度阈值时，进行报警。
电子设备在确定出所包含人体目标的人体高度信息后，可以将所确定的人体高度信息与预设高度阈值进行比较，判断所确定的人体高度信息是否低于预设高度阈值。在一种实现方式中，深度图像中可以包括一个或多个人体目标，当存在多个人体目标时，可以将每一人体目标的人体高度信息分别与预设高度阈值进行比较，判断每一人体目标的人体高度信息是否低于预设高度阈值。
当电子设备判断所确定的人体高度信息低于预设高度阈值时，进行报警。其中，电子设备进行报警时，可以是以输出提示信息的形式进行报警，也可以是以发出提示声音的形式进行报警，也可以是以通过变化屏幕亮度的形式进行报警，等等，凡是可以引起监控人员的注意的方式，均可以作为本申请实施例中的报警形式。
在一种实现方式中，当电子设备确定深度图像中所包含的人体目标为多个时，可以是仅当所包含的人体目标的人体高度信息均低于预设高度阈值时，才会进行报警。
上述预设高度阈值可以是根据作为小孩的人群的身高进行设定的，在一种实现方式中，上述作为小孩的人群可以包括10岁以下的人群。
应用本申请实施例，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的人体高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的人体高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行 关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了监控人员进行人工监控时，可能存在因为人工精力有限或者人为疏忽等因素，而出现监控遗漏的情况，进而造成出现惨剧的情况。
电子设备在确定深度图像中是否包含人体目标时，可以直接利用深度图像中的图像特征，确定深度图像中是否包含人体目标，也可以是利用深度图像对应的点云数据的特征，确定深度图像中是否包含人体目标，这都是可以的。
在一种实现方式中，所述确定所获得的深度图像中是否包含人体目标(S102)的步骤，可以包括：从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
可以理解的是，上述第一预设分类器可以是：基于训练样本以及预设算法训练所得的机器学习模型，其中，上述训练样本可以为：包含人体目标和非人体目标的样本深度图像，即训练上述第一预设分类器所利用的图像特征可以为：上述样本深度图像中的图像特征。对于上述预设算法来说，上述预设算法可以为：包括卷积神经网络相关算法在内的神经网络算法，此时，上述机器学习模型可以为：神经网络模型。或者，上述预设算法可以为：随机森林算法，此时，上述机器学习模型可以为：随机森林分类模型。或者，上述预设算法可以为：支持向量机算法，此时，上述机器学习模型可以为：支持向量机分类模型，等等。
在一种实现方式中，电子设备可以对深度图像进行图像特征提取，例如：利用Canny算子等对深度图像进行边缘特征提取。后续的，将所提取的图像特征输入第一预设分类器中，第一预设分类器对所提取的图像特征进行分类，以确定深度图像中是否包含人体目标。
或者，电子设备中也可以预存有人体图像特征。电子设备从深度图像中提取图像特征后，将所提取的图像特征与预存的人体图像特征进行匹配，当匹配成功时，则可以确定深度图像中包含人体目标。
在另一种实现方式中，所述确定所获得的深度图像中是否包含人体目标(S102)的步骤，可以包括：将所获得的深度图像转换为世界坐标系下的点云数据；确定点云数据中是否包含人体目标对应的点云数据，其中，当判断点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
本申请实施例中，电子设备可以首先将上述深度图像转换为世界坐标系下的点云数据，然后基于上述世界坐标系下的点云数据，确定深度图像中是否包含人体目标。图像采集设备可以利用所包含的深度图像采集子设备采集深度图像，电子设备将深度图像转化为世界坐标系下的点云数据时，可以利用采集该深度图像的图像采集设备中的深度图像采集子设备的参数信息，将深度图像转换为世界坐标系下的点云数据。其中，上述参数信息可以包括深度图像采集子设备的内参和外参，电子设备可以基于张正友标定法对图像采集设备进行标定，从而确定上述参数信息，或者，电子设备可以直接将图像采集设备出厂时所标定的内参，作为本申请实施例中所提的参数信息中的内参，外参可以通过测量及预先标定获得，这都是可以的。在一种实现方式中，所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，可以包括：获得深度图像采集子设备的参数信息，其中，参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用焦距信息、像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，设备坐标系可以为：基于深度图像采集子设备的光心所建立的坐标系；利用安装高度信息以及安装角度信息，将设备坐标系下的点云数据，转化为世界坐标系下的点云数据。
可以理解的是，上述像主点可以为：深度图像采集子设备的光轴与像平面的交点。本申请实施例中，像主点信息可以包含：像主点在深度图像中的二维坐标。其中，利用像主点的二维坐标，可以确定深度图像中每一像素点的二维坐标。
在一种实现方式中，将深度图像中每一像素点的二维坐标(u,v)，转换为设备坐标系中的三维坐标(X C,Y C,Z C)，以获得深度图像的点云数据。其中，上述预设的三维直角坐标系可以为：基于深度图像采集子设备的光心建立的坐标系，坐标转换时所利用公式如下：
Figure PCTCN2018101545-appb-000001
其中，f Dx和f Dy均为深度图像采集子设备的焦距；(u D0,v D0)为上述像主点的二维坐标；Z C为上述深度图像中的像素点(u,v)对应的场景中的点，到深度图像采集子设备的距离信息，即像素点(u,v)的像素值。其中，f Dx表示所确定的上述深度图像采集子设备的x轴方向上的焦距，f Dy表示所确定的上述深度图像采集子设备的y轴方向上的焦距。上述f Dx和f Dy均包含于上述焦距信息中，可以由张正友标定法直接标定确定。上述像主点的二维坐标也可以由张正友标定法直接标定确定。
上述设备坐标系可以为三维直角坐标系。后续的，利用深度图像采集子设备的安装高度信息以及安装角度信息，包括深度图像采集子设备的俯仰角、偏转角，确定上述设备坐标系与世界坐标系的转换关系，进而基于所确定的转换关系将设备坐标系下的点云数据，转化为世界坐标系下的点云数据。
当电子设备将上述深度图像转化为世界坐标系下的点云数据后，可以利用预设的聚类算法对上述点云数据进行聚类，确定出深度图像中所包含的各个目标，并进一步的，确定每一目标是否为人体目标，一种情况中，所述确定点云数据中是否包含人体目标对应的点云数据的步骤，可以包括：对点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
另一种情况中，所述确定点云数据中是否包含人体目标对应的点云数据的步骤，可以包括：对点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，上述预设的聚类算法可以是基于密度的聚类算法，例如：DBSCAN(Density-Based Spatial Clustering of Applications with Noise，具有噪声的基于密度的聚类方法)聚类算法，其中，DBSCAN聚类算法需要 二个参数，分别为扫描半径(eps)和最小包含点数(minPts)。电子设备获得上述点云数据后，利用上述DBSCAN聚类算法从上述点云数据中，任选一个未被访问(unvisited)的点(即未被标记为已访问的点或未被标记为噪声的点)，作为当前点，并确定出与该当前点的距离在eps之内(包括eps)的所有的点，作为附近点；电子设备确定该当前点的附近点的数量是否大于等于minPts，如果电子设备确定该当前点的附近点的数量大于等于minPts，则电子设备将当前点与所确定的附近点确定为一个簇，电子设备将该当前点标记为已访问(visited)；电子设备针对该簇进行递归，以上述方式处理该簇内所有未被标记为已访问(visited)或未被标记为噪声的点，从而对簇进行扩展；如果簇充分地被扩展，即簇内的所有点被标记为已访问；电子设备重新任选一个未被访问的点，作为当前点，执行后续流程。当电子设备确定当前点的附近点的数量小于minPts，则电子设备标记该当前点为噪声；直至点云数据中不包含未被访问的点后，聚类完成。此时，电子设备可以确定出点云数据中所包含的各类点云子数据，后续的，电子设备可以针对每类点云子数据进行处理，以确定每一类点云子数据是否为人体目标对应的点云子数据。
在一种情况中，可以是电子设备将所获得的每一类点云子数据输入第二预设分类器中，以使第二预设分类器对每一类点云子数据进行分类，确定每一类点云子数据是否为人体目标对应的点云子数据。可以理解的是，上述第二预设分类器可以为：基于训练样本以及预设算法训练所得的机器学习模型，其中，上述训练样本可以为：包含人体目标和非人体目标的样本深度图像对应的点云数据。对于上述预设算法来说，上述预设算法可以为：包括卷积神经网络相关算法在内的神经网络算法，此时，上述机器学习模型可以为：神经网络模型。或者，上述预设算法可以为：随机森林算法，此时，上述机器学习模型可以为：随机森林分类模型。或者，上述预设算法可以为：支持向量机算法，此时，上述机器学习模型可以为：支持向量机分类模型，等等。
在另一种情况中，可以是电子设备将所获得的每一类点云子数据以预设的人体模型进行匹配，当匹配成功时，确定点云子数据为人体目标对应的点云子数据，当匹配失败时，确定点云子数据不为人体目标对应的点云子数据。
在一种实现方式中，上述预设的聚类算法还可以为基于网格的聚类算法，例如：STING聚类算法等，也可以是基于划分的聚类算法，例如：K-means聚 类算法等等，本申请实施例并不对能够对点云数据进行聚类的算法进行限定，凡是可以对点云数据进行聚类的算法，均可以应用于本申请实施例中。
在一种实现方式中，可以利用针对电梯轿厢所采集的深度图像对应的色彩图像，确定深度图像中是否包含人体目标，其中，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，可以包括：获得图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；所述确定所获得的深度图像中是否包含人体目标的步骤，可以包括：通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
可以通过人脸识别算法从色彩图像中识别出是否包含人体目标，来确定深度图像中是否包含人体目标，其中，当色彩图像中包含人体目标时，深度图像中包含人体目标；当色彩图像中不包含人体目标时，深度图像中不包含人体目标。
本申请实施例中，上述色彩图像可以为任一色彩模式的图像，例如：RGB(Red Green Blue，红绿蓝)色彩模式的图像，或者YUV色彩模式图像，等等。其中，YUV(也称YCrCb)是一种颜色编码方法，“Y”表示明亮度(Luminance或Luma)，也就是灰阶值；而“U”和“V”表示色度(Chrominance或Chroma)，作用是描述图像色彩及饱和度，用于指定像素点的颜色。“色度”定义了像素点的颜色的两个方面─色调与饱和度，可以分别用Cr和Cb来表示。
在一种情况中，当仅有一个小孩乘坐电梯时，可能出现危险的概率可能会更大些，为了更好的避免上述仅有一个小孩乘坐电梯时，出现危险，当仅有一个小孩乘坐电梯时，监控人员更需要关注并采取相应措施，避免上述独自乘坐电梯的小孩出现危险。在一种实现方式中，如图2所示，所述方法可以包括如下步骤：
S201：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像。
S202：确定所获得的深度图像中是否包含人体目标。
其中，上述S201与图1中所示的S101相同，上述S202与图1中所示的S102 相同。
S203：当确定包含人体目标时，判断所包含人体目标是否为一个。
S204：基于所获得的深度图像，确定所包含人体目标的人体高度信息。
S205：判断所确定的人体高度信息是否低于预设高度阈值。
S206：当判断所确定的人体高度信息低于预设高度阈值时，进行报警。
其中，上述S204与图1中所示的S103相同，上述S205与图1中所示的S104相同，上述S206与图1中所示的S105相同。
在一种实现方式中，当电子设备确定电梯轿厢中所包含的人体目标的人体高度信息均低于预设高度阈值，即确定电梯轿厢中所包含的人体目标均为小孩时，且当人体目标的数量低于预设数量时，需要提醒监控人员进行关注，监控人员针对上述情况采取相应措施，避免上述人体目标即小孩出现危险。在所述进行报警(S105)的步骤之前，所述方法还可以包括：当判断所确定的人体高度信息低于预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
其中，上述确定所包含人体目标的数量的步骤，可以在判断所确定的人体高度信息是否低于预设高度阈值的步骤与进行报警的步骤之间进行，即在判断出所确定的人体高度信息均低于预设高度阈值后，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，进行报警。其中，上述预设数量可以为任一正整数，例如，上述预设数量可以取值为3，即当电子设备确定电梯轿厢内所包含的人体目标的数量低于3，即小孩的数量为2个或1个时，电子设备进行报警。又例如，上述预设数量可以取值为2，即当电子设备确定电梯轿厢内所包含的人体目标的数量低于2，即小孩的数量为1个时，电子设备进行报警。
在一种实现方式中，在所述进行报警(S105)的步骤之后，所述方法还可以包括：控制电梯轿厢移动至指定楼层；当电梯轿厢移动至所述指定楼层后，控制电梯轿厢停止移动；当获得针对电梯轿厢的开门指令后，控制电梯轿厢开启厢门。
本申请实施例中，电子设备进行报警后，可以接收监控人员针对电梯轿厢的移动控制指令，或电子设备进行报警的同时，自动触发移动控制指令等等，这都是可以的。当电子设备获得上述移动控制指令后，电子设备可以控制上述电梯轿厢移动至指定楼层，并控制电梯轿厢停止在上述指定层。
其中，上述指定层可以是预先设置的一层楼层，如一层；或者，可以是在电梯轿厢移动方向上的且与电梯轿厢当前所处位置最近的一层楼层，电梯轿厢当前所处位置可以为：电子设备获得上述移动控制指令时电梯轿厢所处位置。其中，电梯轿厢的移动方向可以包括上升或下降。
当电子设备控制电梯轿厢停止在上述指定层后，可以控制上述电梯轿厢的轿厢门处于关闭状态，直至获得针对电梯轿厢的开门指令后，控制电梯轿厢开启厢门。
相应于上述方法实施例，本申请实施例还提供了一种电梯轿厢监控装置，如图3所示，所述装置可以包括：第一获得模块310，用于当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；第一确定模块320，用于确定所获得的深度图像中是否包含人体目标；第二确定模块330，用于当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；第一判断模块340，用于判断所确定的人体高度信息是否低于预设高度阈值；报警模块350，用于当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
应用本申请实施例，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的人体高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的人体高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了监控人员进行人工监控时，可能存在因为人工精力有限或者人为疏忽等因素， 而出现监控遗漏的情况，进而造成出现惨剧的情况。
在一种实现方式中，所述第一确定模块320，具体用于从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
在一种实现方式中，所述第一确定模块320包括转换单元和确定单元；所述转换单元，用于将所获得的深度图像转换为世界坐标系下的点云数据；所述确定单元，用于确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
在一种实现方式中，所述确定单元，具体用于对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述确定单元，具体用于对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；所述转换单元，具体用于获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
在一种实现方式中，所述第一获得模块310，具体用于获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；所述第一确定模块320，具体用于通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
在一种实现方式中，如图4所示，所述装置还可以包括第二判断模块410； 所述第二判断模块410，用于在所述基于所获得的深度图像，确定所包含人体目标的高度信息之前，当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，触发所述第二确定模块330。
在一种实现方式中，所述装置还可以包括第三确定模块；所述第三确定模块，用于在所述进行报警之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，触发所述报警模块350。
在一种实现方式中，所述装置还可以包括第一控制模块、第二控制模块和第三控制模块；所述第一控制模块，用于在所述进行报警之后，控制所述电梯轿厢移动至指定楼层；所述第二控制模块，用于当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；所述第三控制模块，用于当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
相应于上述方法实施例，本申请实施例还提供一种电梯系统，包括：电梯，以及上述任一电梯轿厢监控装置。
可选地，该电梯系统，包括摄像头，用于采集电梯轿厢内的深度图像，并发送给所述电梯轿厢监控装置。
相应于上述方法实施例，本申请实施例还提供一种电子设备，包括处理器和存储器，其中，所述存储器，用于存放计算机程序；所述处理器，用于执行存储器上所存放的计算机程序时，实现上述任一电梯轿厢监控方法。
相应于上述方法实施例，本发明实施例还提供了一种电子设备，如图5所示，包括处理器510、通信接口520、存储器530和通信总线540，其中，处理器510，通信接口520，存储器530通过通信总线540完成相互间的通信，存储器530，用于存放计算机程序；处理器510，用于执行存储器530上所存放的计算机程序时，实现本申请实施例所提供的任一所述的电梯轿厢监控方法，该方法可以包括如下步骤：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；确定所获得的深度图像中是否包含人体目标；当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；判断所确定的人体高度信息是否低于预设高度阈值；当 判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
应用本申请实施例，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的人体高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的人体高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了监控人员进行人工监控时，可能存在因为人工精力有限或者人为疏忽等因素，而出现监控遗漏的情况，进而造成出现惨剧的情况。
在一种实现方式中，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
在一种实现方式中，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：将所获得的深度图像转换为世界坐标系下的点云数据；确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
在一种实现方式中，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，包括：获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
在一种实现方式中，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，包括：获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；所述确定所获得的深度图像中是否包含人体目标的步骤，包括：通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
在一种实现方式中，所述处理器510还用于实现如下步骤：在所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤之前，当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，执行所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤。
在一种实现方式中，所述处理器510还用于实现如下步骤：在所述进行报警的步骤之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
在一种实现方式中，所述处理器510还用于实现如下步骤：在所述进行报警的步骤之后，控制所述电梯轿厢移动至指定楼层；当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
上述电子设备提到的通信总线可以是外设部件互连标准(Peripheral Component Interconnect，PCI)总线或扩展工业标准结构(Extended Industry Standard Architecture，EISA)总线等。该通信总线可以分为地址总线、数据 总线、控制总线等。为便于表示，图中仅用一条粗线表示，但并不表示仅有一根总线或一种类型的总线。
通信接口用于上述电子设备与其他设备之间的通信。
存储器可以包括随机存取存储器(Random Access Memory，RAM)，也可以包括非易失性存储器(Non-Volatile Memory，NVM)，例如至少一个磁盘存储器。可选的，存储器还可以是至少一个位于远离前述处理器的存储装置。
上述的处理器可以是通用处理器，包括中央处理器(Central Processing Unit，CPU)、网络处理器(Network Processor，NP)等；还可以是数字信号处理器(Digital Signal Processing，DSP)、专用集成电路(Application Specific Integrated Circuit，ASIC)、现场可编程门阵列(Field-Programmable Gate Array，FPGA)或者其他可编程逻辑器件、分立门或者晶体管逻辑器件、分立硬件组件。
相应于上述方法实施例，本申请实施例提供了一种计算机可读存储介质，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现本申请实施例所提供的任一所述的电梯轿厢监控方法，该方法可以包括如下步骤：当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；确定所获得的深度图像中是否包含人体目标；当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；判断所确定的人体高度信息是否低于预设高度阈值；当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
应用本申请实施例，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的人体高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的人体高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了 监控人员进行人工监控时，可能存在因为人工精力有限或者人为疏忽等因素，而出现监控遗漏的情况，进而造成出现惨剧的情况。
在一种实现方式中，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：从所获得的深度图像中提取图像特征；基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
在一种实现方式中，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：将所获得的深度图像转换为世界坐标系下的点云数据；确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
在一种实现方式中，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：对所述点云数据进行聚类，获得各类点云子数据；基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
在一种实现方式中，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，包括：获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
在一种实现方式中，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，包括：获得所述图像采集设备针对电梯轿厢场景所采集的 深度图像，以及所获得的深度图像对应的色彩图像；所述确定所获得的深度图像中是否包含人体目标的步骤，包括：通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
在一种实现方式中，所述计算机程序被处理器执行时还可以实现如下步骤：在所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤之前，当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，执行所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤。
在一种实现方式中，所述计算机程序被处理器执行时还可以实现如下步骤：在所述进行报警的步骤之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
在一种实现方式中，所述计算机程序被处理器执行时还可以实现如下步骤：在所述进行报警的步骤之后，控制所述电梯轿厢移动至指定楼层；当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
相应于上述方法实施例，本申请实施例提供了一种计算机程序产品，当其在计算机上运行时，使得计算机执行上述实施例中任一所述的电梯轿厢监控方法。
本申请实施例提供的方案，当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像，并确定深度图像中是否包含人体目标，当确定包含人体目标时，进一步确定人体目标的人体高度信息，并结合预设高度阈值，判断所包含的人体目标是否为小孩，当判断上述人体目标的人体高度信息低于预设高度阈值时，可确定所包含的人体目标为小孩，且可以确定此时电梯轿厢内出现仅有小孩搭乘的情况，即仅有小孩乘坐电梯的情况。在监控到仅有小孩乘坐电梯的情况下，需要进行报警，以警示监控人员进行关注，后续的，监控人员可以进行相应处理，避免乘坐电梯的小孩出 现危险，以实现对电梯内仅有小孩搭乘的情况的智能监控，节省人力。并且，避免了监控人员进行人工监控时，可能存在因为人工精力有限或者人为疏忽等因素，而出现监控遗漏的情况，进而造成出现惨剧的情况。
需要说明的是，在本文中，诸如第一和第二等之类的关系术语仅仅用来将一个实体或者操作与另一个实体或操作区分开来，而不一定要求或者暗示这些实体或操作之间存在任何这种实际的关系或者顺序。而且，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、方法、物品或者设备不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、方法、物品或者设备所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括所述要素的过程、方法、物品或者设备中还存在另外的相同要素。
本说明书中的各个实施例均采用相关的方式描述，各个实施例之间相同相似的部分互相参见即可，每个实施例重点说明的都是与其他实施例的不同之处。尤其，对于系统实施例而言，由于其基本相似于方法实施例，所以描述的比较简单，相关之处参见方法实施例的部分说明即可。
以上所述仅为本申请的较佳实施例而已，并非用于限定本申请的保护范围。凡在本申请的精神和原则之内所作的任何修改、等同替换、改进等，均包含在本申请的保护范围内。

Claims (34)

  1. 一种电梯轿厢监控方法，其特征在于，所述方法包括：
    当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；
    确定所获得的深度图像中是否包含人体目标；
    当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；
    判断所确定的人体高度信息是否低于预设高度阈值；
    当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
  2. 根据权利要求1所述的方法，其特征在于，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    从所获得的深度图像中提取图像特征；
    基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
  3. 根据权利要求1所述的方法，其特征在于，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    将所获得的深度图像转换为世界坐标系下的点云数据；
    确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
  4. 根据权利要求3所述的方法，其特征在于，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  5. 根据权利要求3所述的方法，其特征在于，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  6. 根据权利要求3所述的方法，其特征在于，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；
    所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，包括：
    获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；
    利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；
    利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
  7. 根据权利要求1所述的方法，其特征在于，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，包括：
    获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；
    所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
  8. 根据权利要求1-7任一项所述的方法，其特征在于，在所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤之前，所述方法还包括：
    当确定包含人体目标时，判断所包含人体目标是否为一个；
    当判断所包含人体目标为一个时，执行所述基于所获得的深度图像，确 定所包含人体目标的高度信息的步骤。
  9. 根据权利要求1-7任一项所述的方法，其特征在于，在所述进行报警的步骤之前，所述方法还包括：
    当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；
    当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
  10. 根据权利要求1-7任一项所述的方法，其特征在于，在所述进行报警的步骤之后，所述方法还包括：
    控制所述电梯轿厢移动至指定楼层；
    当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；
    当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
  11. 一种电梯轿厢监控装置，其特征在于，所述装置包括：
    第一获得模块，用于当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；
    第一确定模块，用于确定所获得的深度图像中是否包含人体目标；
    第二确定模块，用于当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；
    第一判断模块，用于判断所确定的人体高度信息是否低于预设高度阈值；
    报警模块，用于当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
  12. 根据权利要求11所述的装置，其特征在于，所述第一确定模块，具体用于
    从所获得的深度图像中提取图像特征；
    基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中 是否包含人体目标。
  13. 根据权利要求11所述的装置，其特征在于，所述第一确定模块包括转换单元和确定单元；
    所述转换单元，用于将所获得的深度图像转换为世界坐标系下的点云数据；
    所述确定单元，用于确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
  14. 根据权利要求13所述的装置，其特征在于，所述确定单元，具体用于
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  15. 根据权利要求13所述的装置，其特征在于，所述确定单元，具体用于
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  16. 根据权利要求13所述的装置，其特征在于，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；
    所述转换单元，具体用于
    获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；
    利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；
    利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
  17. 根据权利要求11所述的装置，其特征在于，所述第一获得模块，具体用于
    获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；
    所述第一确定模块，具体用于
    通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
  18. 根据权利要求11-17任一项所述的装置，其特征在于，所述装置还包括第二判断模块；
    所述第二判断模块，用于在所述基于所获得的深度图像，确定所包含人体目标的高度信息之前，当确定包含人体目标时，判断所包含人体目标是否为一个；当判断所包含人体目标为一个时，触发所述第二确定模块。
  19. 根据权利要求11-17任一项所述的装置，其特征在于，所述装置还包括第三确定模块；
    所述第三确定模块，用于在所述进行报警之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；
    当所确定的所包含人体目标的数量低于预设数量时，触发所述报警模块。
  20. 根据权利要求11-17任一项所述的装置，其特征在于，所述装置还包括第一控制模块、第二控制模块和第三控制模块；
    所述第一控制模块，用于在所述进行报警之后，控制所述电梯轿厢移动至指定楼层；
    所述第二控制模块，用于当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；
    所述第三控制模块，用于当获得针对所述电梯轿厢的开门指令后，控制 所述电梯轿厢开启厢门。
  21. 一种电子设备，其特征在于，包括处理器和存储器，其中，所述存储器，用于存放计算机程序；
    所述处理器，用于执行存储器上所存放的计算机程序时，实现如下步骤：
    当检测到电梯轿厢门关闭时，获得图像采集设备针对电梯轿厢场景所采集的深度图像；
    确定所获得的深度图像中是否包含人体目标；
    当确定包含人体目标时，基于所获得的深度图像，确定所包含人体目标的人体高度信息；
    判断所确定的人体高度信息是否低于预设高度阈值；
    当判断所确定的人体高度信息低于所述预设高度阈值时，进行报警。
  22. 根据权利要求21所述的电子设备，其特征在于，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    从所获得的深度图像中提取图像特征；
    基于所提取的图像特征以及第一预设分类器，确定所获得的深度图像中是否包含人体目标。
  23. 根据权利要求21所述的电子设备，其特征在于，所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    将所获得的深度图像转换为世界坐标系下的点云数据；
    确定所述点云数据中是否包含人体目标对应的点云数据，其中，当判断所述点云数据中包含人体目标对应的点云数据时，表征所获得的深度图像中包含人体目标。
  24. 根据权利要求23所述的电子设备，其特征在于，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及第二预设分类器，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  25. 根据权利要求23所述的电子设备，其特征在于，所述确定所述点云数据中是否包含人体目标对应的点云数据的步骤，包括：
    对所述点云数据进行聚类，获得各类点云子数据；
    基于所获得的各类点云子数据以及预设人体模型，确定所获得的各类点云子数据中是否包含人体目标对应的点云子数据。
  26. 根据权利要求23所述的电子设备，其特征在于，所述图像采集设备利用所包含的深度图像采集子设备采集深度图像；
    所述将所获得的深度图像转换为世界坐标系下的点云数据的步骤，包括：
    获得所述深度图像采集子设备的参数信息，其中，所述参数信息中包含：焦距信息、像主点信息、安装高度信息以及安装角度信息；
    利用所述焦距信息、所述像主点信息，将所获得的深度图像转换为设备坐标系下的点云数据，其中，所述设备坐标系为：基于所述深度图像采集子设备的光心所建立的坐标系；
    利用所述安装高度信息以及所述安装角度信息，将所述设备坐标系下的点云数据，转化为所述世界坐标系下的点云数据。
  27. 根据权利要求21所述的电子设备，其特征在于，所述获得图像采集设备针对电梯轿厢场景所采集的深度图像的步骤，包括：
    获得所述图像采集设备针对电梯轿厢场景所采集的深度图像，以及所获得的深度图像对应的色彩图像；
    所述确定所获得的深度图像中是否包含人体目标的步骤，包括：
    通过识别所述色彩图像中是否包含人体目标，确定获得的深度图像中是否包含人体目标。
  28. 根据权利要求21-27任一项所述的电子设备，其特征在于，所述处理器还用于实现如下步骤：
    在所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤之前，在所述基于所获得的深度图像，确定所包含人体目标的高度信息之前，当确定包含人体目标时，判断所包含人体目标是否为一个；
    当判断所包含人体目标为一个时，执行所述基于所获得的深度图像，确定所包含人体目标的高度信息的步骤。
  29. 根据权利要求21-27任一项所述的电子设备，其特征在于，所述处理器还用于实现如下步骤：
    在所述进行报警的步骤之前，当判断所确定的人体高度信息低于所述预设高度阈值时，确定所包含人体目标的数量；
    当所确定的所包含人体目标的数量低于预设数量时，执行所述进行报警的步骤。
  30. 根据权利要求21-27任一项所述的电子设备，其特征在于，所述处理器还用于实现如下步骤：
    在所述进行报警的步骤之后，控制所述电梯轿厢移动至指定楼层；
    当所述电梯轿厢移动至所述指定楼层后，控制所述电梯轿厢停止移动；
    当获得针对所述电梯轿厢的开门指令后，控制所述电梯轿厢开启厢门。
  31. 一种电梯系统，其特征在于，包括：电梯，以及如权利要求11-20中任一所述的电梯轿厢监控装置。
  32. 根据权利要求31所述的电梯系统，其特征在于，包括：
    摄像头，用于采集电梯轿厢内的深度图像，并发送给所述电梯轿厢监控装置。
  33. 一种计算机可读存储介质，其特征在于，所述计算机可读存储介质内存储有计算机程序，所述计算机程序被处理器执行时实现权利要求1-10任一所述的电梯轿厢监控方法步骤。
  34. 一种计算机程序产品，其特征在于，当其在计算机上运行时，使得计算机执行权利要求1-10任一所述的电梯轿厢监控方法步骤。
PCT/CN2018/101545 2017-09-13 2018-08-21 一种电梯轿厢监控方法、装置及系统 WO2019052318A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201710822052.0A CN109484935B (zh) 2017-09-13 2017-09-13 一种电梯轿厢监控方法、装置及系统
CN201710822052.0 2017-09-13

Publications (1)

Publication Number Publication Date
WO2019052318A1 true WO2019052318A1 (zh) 2019-03-21

Family

ID=65688677

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/101545 WO2019052318A1 (zh) 2017-09-13 2018-08-21 一种电梯轿厢监控方法、装置及系统

Country Status (2)

Country Link
CN (1) CN109484935B (zh)
WO (1) WO2019052318A1 (zh)

Cited By (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112693981A (zh) * 2020-12-17 2021-04-23 日立楼宇技术（广州）有限公司 一种触摸操纵箱的调整方法、装置、设备及存储介质
CN113104687A (zh) * 2021-04-14 2021-07-13 合肥工业大学 基于计算机视觉识别判断电梯空间状态的方法及装置
CN113269111A (zh) * 2021-06-03 2021-08-17 昆山杜克大学 一种基于视频监控的电梯异常行为检测方法及系统
CN113838097A (zh) * 2021-09-29 2021-12-24 成都新潮传媒集团有限公司 摄像机镜头角度偏离检测方法、装置及存储介质
CN114155557A (zh) * 2021-12-07 2022-03-08 美的集团(上海)有限公司 定位方法、装置、机器人及计算机可读存储介质
CN115909321A (zh) * 2023-03-08 2023-04-04 成都睿瞳科技有限责任公司 一种用于电梯轿厢的识别方法、系统和存储介质
CN116137074A (zh) * 2023-02-22 2023-05-19 常熟理工学院 电梯轿厢内乘客打斗行为的自动检测方法和系统
WO2023245402A1 (en) * 2022-06-21 2023-12-28 Kone Corporation A method and a triggering system for triggering an acquisition of image data of an empty elevator car

Families Citing this family (9)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111031278B (zh) * 2019-11-25 2021-02-05 广州恒龙信息技术有限公司 一种基于结构光和tof的监控方法和系统
CN111689327A (zh) * 2020-06-19 2020-09-22 迅达（中国）电梯有限公司 电梯轿厢控制系统
CN114063572B (zh) * 2020-07-31 2024-05-31 北京熵行科技有限公司 一种无感知智能设备控制方法、电子设备及控制系统
CN112184737B (zh) * 2020-10-10 2024-06-14 上海阅目科技有限公司 一种基于图像分析的儿童独乘电梯检测及报警系统
CN112257570B (zh) * 2020-10-20 2021-07-27 江苏濠汉信息技术有限公司 基于视觉分析检测施工人员安全帽未佩戴的方法及装置
CN112529953B (zh) * 2020-12-17 2022-05-03 深圳市普渡科技有限公司 电梯的空间状态判断方法、装置及存储介质
CN113361410B (zh) * 2021-06-07 2022-08-26 上海数川数据科技有限公司 一种基于网格聚类的小孩检测框过滤算法
CN113492405B (zh) * 2021-06-17 2022-08-02 达闼机器人股份有限公司 机器人开门方法、装置、存储介质及电子设备
CN114180431B (zh) * 2021-12-22 2023-12-15 佛山市和众电梯技术有限公司 一种轿厢电梯安全运行监控系统

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102657532A (zh) * 2012-05-04 2012-09-12 深圳泰山在线科技有限公司 基于人体姿态识别的身高测量方法及装置
CN104176570A (zh) * 2013-05-22 2014-12-03 日立电梯（中国）有限公司 一种电梯的控制方法及系统
CN105129550A (zh) * 2015-08-03 2015-12-09 陈思 防止未成年人单独乘坐的电梯安全控制平台
WO2016068869A1 (en) * 2014-10-28 2016-05-06 Hewlett-Packard Development Company, L.P. Three dimensional object recognition
CN106315324A (zh) * 2015-06-16 2017-01-11 奥的斯电梯公司 一种能够监测儿童使用的电梯系统及其控制方法
CN206486130U (zh) * 2017-01-19 2017-09-12 北京北迅电梯公司 电梯预警系统

Family Cites Families (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US5274714A (en) * 1990-06-04 1993-12-28 Neuristics, Inc. Method and apparatus for determining and organizing feature vectors for neural network recognition
CN100595121C (zh) * 2007-12-10 2010-03-24 中国科学院计算技术研究所 一种目标场所优化调度方法及系统
CN103413145B (zh) * 2013-08-23 2016-09-21 南京理工大学 基于深度图像的关节点定位方法
CN203728342U (zh) * 2014-03-05 2014-07-23 中国石油大学(华东) 一种电梯防误停系统
CN103971135A (zh) * 2014-05-05 2014-08-06 中国民航大学 基于头肩深度信息特征的人体目标检测方法
CN104598890B (zh) * 2015-01-30 2017-07-28 南京邮电大学 一种基于rgb‑d视频的人体行为识别方法
CN104751146B (zh) * 2015-04-13 2018-04-10 中国科学技术大学 一种基于3d点云图像的室内人体检测方法
CN105923481B (zh) * 2016-06-30 2017-12-19 爱默生电梯有限公司 一种防止幼儿误入危险的升降电梯

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102657532A (zh) * 2012-05-04 2012-09-12 深圳泰山在线科技有限公司 基于人体姿态识别的身高测量方法及装置
CN104176570A (zh) * 2013-05-22 2014-12-03 日立电梯（中国）有限公司 一种电梯的控制方法及系统
WO2016068869A1 (en) * 2014-10-28 2016-05-06 Hewlett-Packard Development Company, L.P. Three dimensional object recognition
CN106315324A (zh) * 2015-06-16 2017-01-11 奥的斯电梯公司 一种能够监测儿童使用的电梯系统及其控制方法
CN105129550A (zh) * 2015-08-03 2015-12-09 陈思 防止未成年人单独乘坐的电梯安全控制平台
CN206486130U (zh) * 2017-01-19 2017-09-12 北京北迅电梯公司 电梯预警系统

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112693981A (zh) * 2020-12-17 2021-04-23 日立楼宇技术（广州）有限公司 一种触摸操纵箱的调整方法、装置、设备及存储介质
CN112693981B (zh) * 2020-12-17 2023-04-18 日立楼宇技术（广州）有限公司 一种触摸操纵箱的调整方法、装置、设备及存储介质
CN113104687A (zh) * 2021-04-14 2021-07-13 合肥工业大学 基于计算机视觉识别判断电梯空间状态的方法及装置
CN113269111A (zh) * 2021-06-03 2021-08-17 昆山杜克大学 一种基于视频监控的电梯异常行为检测方法及系统
CN113269111B (zh) * 2021-06-03 2024-04-05 昆山杜克大学 一种基于视频监控的电梯异常行为检测方法及系统
CN113838097A (zh) * 2021-09-29 2021-12-24 成都新潮传媒集团有限公司 摄像机镜头角度偏离检测方法、装置及存储介质
CN113838097B (zh) * 2021-09-29 2024-01-09 成都新潮传媒集团有限公司 摄像机镜头角度偏离检测方法、装置及存储介质
CN114155557A (zh) * 2021-12-07 2022-03-08 美的集团(上海)有限公司 定位方法、装置、机器人及计算机可读存储介质
CN114155557B (zh) * 2021-12-07 2022-12-23 美的集团(上海)有限公司 定位方法、装置、机器人及计算机可读存储介质
WO2023245402A1 (en) * 2022-06-21 2023-12-28 Kone Corporation A method and a triggering system for triggering an acquisition of image data of an empty elevator car
CN116137074A (zh) * 2023-02-22 2023-05-19 常熟理工学院 电梯轿厢内乘客打斗行为的自动检测方法和系统
CN115909321A (zh) * 2023-03-08 2023-04-04 成都睿瞳科技有限责任公司 一种用于电梯轿厢的识别方法、系统和存储介质

Also Published As

Publication number Publication date
CN109484935B (zh) 2020-11-20
CN109484935A (zh) 2019-03-19

Similar Documents

Publication Publication Date Title
WO2019052318A1 (zh) 一种电梯轿厢监控方法、装置及系统
WO2020078229A1 (zh) 目标对象的识别方法和装置、存储介质、电子装置
WO2021120867A1 (zh) 高空抛物的监测方法、装置、电子设备及存储介质
JP6710426B2 (ja) 障害物検出方法及び装置
WO2017190574A1 (zh) 一种基于聚合通道特征的快速行人检测方法
WO2019223586A1 (zh) 车位使用状况的检测方法、装置、电子设备及存储介质
US20150363670A1 (en) Image recognition method and camera system
WO2015184764A1 (zh) 行人检测方法及装置
KR20210149848A (ko) 피부질 검출 방법, 피부질 등급 분류 방법, 피부질 검출 장치, 전자 기기 및 저장 매체
AU2017233723A1 (en) System and method for training object classifier by machine learning
WO2019033572A1 (zh) 人脸遮挡检测方法、装置及存储介质
WO2023273219A1 (zh) 玻璃幕墙开启窗开启状态检测方法、装置、设备及介质
CN110009650B (zh) 一种扶梯扶手边界区域越界检测方法与系统
CN103065163B (zh) 一种基于静态图片的快速目标检测识别系统及方法
CN106991821A (zh) 违章车辆用手持移动终端数据采集系统
WO2021204267A1 (zh) 身份识别
JP2020518879A (ja) 検出システム、検出装置およびその方法
WO2023138163A1 (zh) 一种基于深度图像修复的室内移动机器人玻璃检测与地图更新方法
DE112020001706T5 (de) Personenerkennungsvorrichtung und personenerkennungsverfahren
WO2020010620A1 (zh) 波浪识别方法、装置、计算机可读存储介质和无人飞行器
CN114332925A (zh) 电梯内宠物检测方法、系统、装置及计算机可读存储介质
CN112507767B (zh) 脸部辨识方法及其相关电脑系统
JP2002282210A (ja) 視線検出方法及びその装置
WO2019223763A1 (zh) 图像检测方法、装置、设备及介质，图案化控制系统及方法
CN108460357B (zh) 一种基于图像识别的开窗报警检测系统及方法

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: 18855271

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18855271

Country of ref document: EP

Kind code of ref document: A1