Intelligent controller and elevator

Abstract

The invention relates to an intelligent controller and an elevator, the intelligent controller comprises a control unit, and a first image acquisition unit and a second image acquisition unit which are electrically connected with the control unit, and the first image acquisition unit and the second image acquisition unit are used for feeding back image data to the control unit; the control unit is used for analyzing according to the obtained image data to obtain a newly-added passenger object and an associated feature object belonging to the newly-added passenger object and issuing a control instruction according to the type of the associated feature object, and the control instruction comprises running, stopping and warning reminding. The elevator comprises the intelligent controller. According to the intelligent controller and the elevator, the recognition rate of the battery of the electric vehicle is increased in the mode of collecting the images at the two positions and analyzing the collected images, so that pre-discovery of dangerous factors is achieved, and the safety of the elevator in the running process is improved.

Description

Technical Field

[0001] The present application relates to the technical field of intelligent controllers, and in particular to an intelligent controller and an elevator. Background Art

[0002] The elevator controller is the core part of the elevator system, responsible for managing and coordinating the operation of the elevator to ensure safe and efficient transportation of passengers. It receives signals from inside and outside the elevator (such as floor requests, elevator position, load, etc.) to determine the elevator's running direction, stop floors, and door opening and closing operations.

[0003] The main functions of the elevator controller include dispatching control, operation control, positioning, door opening and closing control, and safety protection. An important aspect of safety protection is to identify dangerous factors in the car during the entry and exit of personnel and operation. For example, electric vehicles are currently strictly prohibited from entering elevators. One of the main ways to achieve safety control is through video monitoring and data analysis.

[0004] In terms of electric vehicle bans, the current recognition rate is relatively high, but in actual use, it is found that some electric vehicles have detachable batteries, and some people still carry the batteries into elevators after removing them, which creates certain risks. The electric vehicle batteries in this scenario have problems such as inconsistent shapes and easy obstruction, which increases the difficulty of recognition, and the risk of battery loss of control will further increase. This is because there is no external obstruction, and the spread speed of battery loss of control will be significantly higher than the spread speed of loss of control when the battery is inside the electric vehicle. Summary of the invention

[0005] The present application provides an intelligent controller and an elevator, which improve the recognition rate of electric vehicle batteries by collecting images at dual positions and analyzing the collected images, so as to achieve pre-discovery of dangerous factors and improve the safety of the elevator operation process.

[0006] The above-mentioned purpose of the present application is achieved through the following technical solutions:

[0007] In a first aspect, the present application provides an intelligent controller, comprising:

[0008] Control unit;

[0009] A first image acquisition unit and a second image acquisition unit, wherein the first image acquisition unit and the second image acquisition unit are both electrically connected to the control unit;

[0010] The first image acquisition unit and the second image acquisition unit are used to feed back image data to the control unit;

[0011] The control unit is used to obtain a newly added passenger object and an associated feature object belonging to the newly added passenger object according to the obtained image data analysis, and issue a control instruction according to the type of the associated feature object;

[0012] Among them, control instructions include run, stop and warning reminders.

[0013] In a possible implementation manner of the first aspect, the control unit obtains the newly added passenger object and the associated feature object belonging to the newly added passenger object according to the obtained image data analysis, including:

[0014] In response to an opening instruction of the elevator car, a first orientation image and a second orientation image are obtained, wherein both the first orientation image and the second orientation image point to a door opening area;

[0015] identifying a passenger object included in the first orientation image and a passenger object included in the second orientation image;

[0016] Associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain a newly added passenger object;

[0017] Determine the associated feature objects that belong to the newly added passenger object.

[0018] In a possible implementation manner of the first aspect, identifying the passenger object includes:

[0019] Acquire a first reference image and a second reference image in a time series, wherein the generation time of the first reference image is earlier than the generation time of the second reference image;

[0020] Comparing the first reference image with the second reference image, and determining a difference area on the first reference image or the second reference image;

[0021] It is determined whether there is a passenger object in the distinguished area, and when the passenger object exists in the distinguished area, the passenger object is marked.

[0022] In a possible implementation manner of the first aspect, when determining whether there is a passenger object in the distinguishing area, the determination is performed using a facial recognition method, and the determination using the facial recognition method includes:

[0023] determining the contours of the distinguishing regions and searching for facial features within the contours;

[0024] Find facial details among facial features;

[0025] Establish an analysis line based on facial detail features and use the pixel points on the analysis line to construct an analysis series;

[0026] Analyze the analytical sequence and obtain the characteristics of the analytical sequence;

[0027] It is determined whether there is a passenger object in the distinguished area based on the characteristics of the analysis sequence.

[0028] In a possible implementation manner of the first aspect, parsing the analysis sequence and obtaining the characteristics of the analysis sequence includes:

[0029] Calculate the difference sequence or quadratic difference sequence of the analytical sequence;

[0030] Determine a slope change region on the difference sequence or the quadratic difference sequence, wherein the number of the slope change regions is multiple;

[0031] Calculate the distance between adjacent slope change areas and record it as the characteristic distance;

[0032] According to the characteristic distance, the reference point of the analysis line is adjusted and the slope change area is calculated again to make the characteristic distance approach the characteristic distance model;

[0033] When the distance between two slope change regions satisfies the characteristic distance model, other slope change regions between the two slope change regions are deleted.

[0034] In a possible implementation manner of the first aspect, associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain the newly added passenger object includes:

[0035] constructing a first analysis grid using passenger objects included in the first orientation image and constructing a second analysis grid using passenger objects included in the second orientation image;

[0036] Comparing the similarity between the first analysis grid and the second analysis grid to obtain a similarity comparison result;

[0037] adjusting the second analysis grid according to the similarity comparison result, and determining the distinguishing grid points existing in the second analysis grid;

[0038] The passenger object corresponding to the distinguished grid point is used as the newly added passenger object.

[0039] In a possible implementation manner of the first aspect, determining the associated feature object belonging to the newly added passenger object includes:

[0040] Establish a simplified model of the newly added passenger object, which includes the head, torso and limbs;

[0041] The suspected area was established based on the extremities of the simplified model;

[0042] Search for associated feature objects in the suspected area. When an associated feature object exists in the suspected area, bind the found associated feature object to the newly added passenger object.

[0043] In a possible implementation manner of the first aspect, searching for the associated feature object in the suspected region includes:

[0044] Dividing the suspected region by color to obtain a plurality of first suspected sub-regions, wherein the color of pixels in each first suspected sub-region tends to be consistent;

[0045] Use the shape to screen the first suspicious sub-region once to obtain the second suspicious sub-region;

[0046] Pick up contour features in the second suspicious sub-region and establish a feature surface based on the contour features;

[0047] Pick up charging features on the feature surface;

[0048] Perform voting calculation on the contour features and the charging features to obtain calculation results;

[0049] The type of the associated feature object is given according to the calculation result.

[0050] In a second aspect, the present application provides an elevator, comprising an intelligent controller as described in the first aspect and any implementation of the first aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0051] Figure 1 It is a schematic diagram of the appearance structure of an elevator car provided in this application.

[0052] Figure 2 This is a schematic diagram of the deployment of an intelligent controller inside an elevator car provided by the present application.

[0053] Figure 3 This is a schematic diagram provided by this application in which there are new personnel.

[0054] Figure 4 This is a schematic diagram of the principle of identifying passenger objects by comparison provided by the present application.

[0055] Figure 5 This is a schematic diagram of the principle of obtaining distinguishing grid points provided by the present application.

[0056] Figure 6 This is a schematic diagram of a principle for obtaining a suspected area provided by the present application.

[0057] In the figure, 1 is a control unit, 2 is a first image acquisition unit, and 3 is a second image acquisition unit. DETAILED DESCRIPTION

[0058] The technical solution in this application is further described in detail below in conjunction with the accompanying drawings.

[0059] This application discloses an intelligent controller, see Figure 1 and Figure 2 In some examples, the intelligent controller disclosed in the present application includes a control unit 1, a first image acquisition unit 2 and a second image acquisition unit 3. The first image acquisition unit 2 and the second image acquisition unit 3 are both installed on the top surface of the interior space of the elevator car. The control unit 1 can be installed on the top surface of the interior space of the elevator car or at other locations on the elevator car. The present application does not limit the installation position of the control unit 1.

[0060] First, the control unit 1 is introduced. The control unit 1 includes a core processor, a sensor module, a communication module, a storage module and a power management module.

[0061] The core processor can be an MCU (microcontroller), MPU (microprocessor), DSP (digital signal processor) and FPGA (field programmable gate array), etc., and its function is to process data. The sensor module mainly includes temperature sensor, humidity sensor, acceleration sensor and gyroscope, etc., and its main function is to collect relevant parameters and posture detection during the operation of the elevator car. The communication module is mainly responsible for data communication with the host computer and the cloud; the storage module is mainly responsible for storing program code, configuration parameters, operation data and historical records; the power management module is responsible for providing a stable power supply for the core processor and other modules.

[0062] The first image acquisition unit 2 and the second image acquisition unit 3 in the present application both use cameras, and data communication between the first image acquisition unit 2 and the second image acquisition unit 3 and the control unit 1 is achieved through USB, HDMI, Ethernet, Wi-Fi, Bluetooth, ZigBee and the like.

[0063] The control unit 1 is used to obtain a new passenger object and an associated feature object belonging to the new passenger object based on the obtained image data analysis. Here, the new passenger object refers to the newly added passengers inside the elevator car, and the associated feature object belonging to the new passenger object refers to the items carried by the newly added passengers inside the elevator car.

[0064] After the items carried by the newly added passengers are determined, control instructions will be issued accordingly. There are three types of control instructions, namely, run, stop and warning reminder. The corresponding scenario for run is that the items carried by the newly added passengers meet the requirements, and the corresponding scenario for stop is that the items carried by the newly added passengers do not meet the requirements. Warning reminder and stop instructions appear at the same time. Generally speaking, the warning reminder is a voice reminder.

[0065] The specific steps for the control unit 1 to obtain the newly added passenger object and the associated feature object belonging to the newly added passenger object according to the obtained image data analysis are as follows:

[0066] S101, in response to an opening instruction of an elevator car, obtaining a first orientation image and a second orientation image, wherein both the first orientation image and the second orientation image point to a door opening area;

[0067] S102, identifying a passenger object included in the first orientation image and a passenger object included in the second orientation image;

[0068] S103, associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain a newly added passenger object;

[0069] S104: Determine the associated feature object belonging to the newly added passenger object.

[0070] Combined with the content mentioned in the background technology, it can be seen that the above steps are mainly used to solve the problem of electric vehicle rechargeable battery identification. There are two situations in the elevator: a small number of passengers and a large number of passengers. In order to improve the recognition rate, the present application uses the first image acquisition unit 2 and the second image acquisition unit 3 to acquire images, that is, the first orientation image and the second orientation image in step S101.

[0071] The first image acquisition unit 2 generates a first orientation image, and the second image acquisition unit 3 generates a second orientation image.

[0072] The first image acquisition unit 2 and the second image acquisition unit 3 are both arranged tilted, such as Figure 2 As shown, Figure 2 The dotted line in represents the field of view of the image sensor.

[0073] In step S101, in response to the opening instruction of the elevator car, the intelligent controller obtains a first orientation image and a second orientation image, both of which point to the door opening area, that is, when the elevator car is opened, the first image acquisition unit 2 and the second image acquisition unit 3 will act simultaneously to generate a first orientation image and a second orientation image.

[0074] In step S102, firstly, the passenger objects included in the first orientation image and the passenger objects included in the second orientation image are identified, where the passenger objects refer to passengers in the elevator car or about to enter the elevator car.

[0075] In step S103, the passenger object included in the first orientation image and the passenger object included in the second orientation image are associated to obtain a new passenger object, and then in step S104, an associated feature object belonging to the new passenger object is determined, where the associated feature object refers to the electric vehicle rechargeable battery.

[0076] Then, any one of operation, stop and voice reminder is issued according to the result of step S104. The corresponding situations of operation, stop and voice reminder are as follows:

[0077] The corresponding case of operation is that there is no associated feature object;

[0078] The corresponding situation of stop is the existence of associated feature objects. The warning reminder is used in conjunction with the stop, which means that passengers are prompted to remove the associated feature objects through voice reminders.

[0079] In general, the intelligent controller provided in this application, for electric vehicle charging batteries, first uses the method of determining new personnel to determine whether there is a potential risk, such as Figure 3 As shown, only the newly added persons are identified.

[0080] Because the existing people in the elevator car have already been screened, there is no need to repeat the screening. The advantage of this method is that it can quickly lock the screening range and increase the screening speed. Figure 3 It can also be seen that when the number of people increases, the probability of the image sensor on the left recognizing the electric vehicle charging battery will be greatly reduced. One of the main reasons is the occlusion problem caused by height.

[0081] In some examples, the specific way to identify the passenger object is:

[0082] S201, acquiring a first reference image and a second reference image in a time series, wherein the generation time of the first reference image is earlier than the generation time of the second reference image;

[0083] S202, comparing the first reference image and the second reference image, and determining a difference area on the first reference image or the second reference image;

[0084] S203, determining whether there is a passenger object in the distinguished area, and when there is a passenger object in the distinguished area, marking the passenger object.

[0085] The contents in step S201 to step S203 are determined by comparison, see Figure 4 Specifically, a first reference image and a second reference image are first obtained, the difference between the first reference image and the second reference image is that the generation time is different, and then the first reference image and the second reference image are compared to determine the difference area on the first reference image or the second reference image, where the difference area refers to the different area in the two photos.

[0086] Finally, it is determined whether there is a passenger object in the distinguished area, and when there is a passenger object in the distinguished area, the passenger object is marked.

[0087] Regarding the method of determining the passenger object, in some possible implementations, a trained portrait recognition neural network is used. The advantage of finding the distinguishing area first here is that there is advertising content in the elevator car. If there is a portrait in the advertising content, it will interfere with the subsequent portrait recognition neural network work. This is because the portrait in the advertising content is also displayed in a flat display manner in the first reference image and the second reference image, which is the same as the display method of the passengers in the elevator car on the first reference image and the second reference image.

[0088] When the different areas are used for preliminary screening, this interference factor can be removed because the portraits included in the advertisement will be eliminated when the different areas are initially screened.

[0089] In addition, for the distinguished areas, the method of finding associated feature objects can also be used for processing. The associated feature objects here refer to electric vehicle rechargeable batteries. This is because if the passenger object uses methods such as occlusion and hiding to process the electric vehicle rechargeable battery, the electric vehicle rechargeable battery will not be identified. However, when the occlusion and hiding methods are removed, there is a certain possibility of finding the processed (occlusion and hiding, etc.) electric vehicle rechargeable battery.

[0090] Of course, the present application also provides another method for determining whether there is a passenger object in the distinguishing area, which is to use facial recognition to determine, the specific method is as follows:

[0091] S301, determining the contour of the distinguished area and searching for facial features inside the contour;

[0092] S302, finding facial detail features in facial features;

[0093] S303, establishing an analysis line based on facial detail features and constructing an analysis series using pixel points on the analysis line;

[0094] S304, analyzing the analysis sequence and obtaining the characteristics of the analysis sequence;

[0095] S305: Determine whether there is a passenger object in the distinguished area according to the characteristics of the analysis sequence.

[0096] A fast processing method is used in steps S301 to S305. This is because for the technical solution described in this application, accurate face comparison and recognition is not required, but fast recognition is required. Therefore, there is no need to perform comparison and recognition of the entire face, but only the ability to identify the passenger object and perform a simple comparison of the passenger object is required.

[0097] In step S301 to step S305, first determine the outline of the distinguished area and search for facial features within the outline, then find facial detail features in the facial features. The facial detail features here generally use eyes, nose and mouth. If the facial features include a mask at this time, the eyes need to be used as facial detail features.

[0098] If the facial details cannot be obtained at this time, the head and the area immediately adjacent to the head are used as an overall feature. The specific method is to judge by the outer contour. Here, the trained neural network is still used for recognition processing. However, it should be noted that this method is only used when the facial details cannot be used, because its recognition accuracy is lower than that of using facial details for recognition.

[0099] Then, an analysis line is established based on the facial detail features and an analysis sequence is constructed using the pixels on the analysis line. The analysis sequence is then parsed and the features of the analysis sequence are obtained. Finally, it is determined whether there is a passenger object in the distinguished area based on the features of the analysis sequence.

[0100] The way of establishing the analysis line based on the facial detail features is to set a straight line in the area including the facial detail features, and the straight line passes through the facial detail features.

[0101] The analysis series is constructed by using the pixel points on the analysis line. The positions of the pixel points on the analysis line are sorted as serial numbers, and the pixel values ​​of the pixel points are used as numerical values ​​to obtain the analysis series.

[0102] The specific method of analyzing the analysis sequence and obtaining the characteristics of the analysis sequence is to calculate the difference sequence or quadratic difference sequence of the analysis sequence. The slope of the curve corresponding to the difference sequence or quadratic difference sequence will change synchronously, as follows:

[0103] Calculate the difference sequence or quadratic difference sequence of the analytical sequence;

[0104] Determine a slope change region on the difference sequence or the quadratic difference sequence, wherein the number of the slope change regions is multiple;

[0105] Calculate the distance between adjacent slope change areas and record it as the characteristic distance;

[0106] According to the characteristic distance, the reference point of the analysis line is adjusted and the slope change area is calculated again to make the characteristic distance approach the characteristic distance model;

[0107] When the distance between two slope change regions satisfies the characteristic distance model, other slope change regions between the two slope change regions are deleted.

[0108] For example, the slope change is set to be greater than or equal to an inherent setting parameter, and the area corresponding to the slope change (such as the starting point or the end point) is used as a feature point. After all the feature points are obtained, the order of the feature points and the distance between adjacent feature points are recorded. The distance between adjacent feature points is the characteristic of the analysis series.

[0109] For the features of the analysis series, when they meet certain requirements, it indicates whether there is a passenger object in the distinguished area. Here, the eyes are used as an example. The distance between the corners of the eyes and the pupils is within a certain range, that is, standard features can be obtained through statistical analysis. The standard features are generally a range. If the features of the analysis series are within this range, it indicates that there is a passenger object in the distinguished area.

[0110] In some possible implementations, there are multiple analysis lines, which may be in parallel or intersecting forms. The purpose of adding analysis lines is to improve the accuracy of judgment.

[0111] In some examples, a method of associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain the newly added passenger object is:

[0112] S401, constructing a first analysis grid using passenger objects included in a first orientation image and constructing a second analysis grid using passenger objects included in a second orientation image;

[0113] S402, comparing the similarity between the first analysis grid and the second analysis grid to obtain a similarity comparison result;

[0114] S403, adjusting the second analysis grid according to the similarity comparison result, and determining the distinguishing grid points existing in the second analysis grid;

[0115] S404: The passenger object corresponding to the distinguished grid point is used as a newly added passenger object.

[0116] See also Figure 5 In step S401 to step S404, firstly, a first analysis grid is constructed using the passenger objects included in the first orientation image and a second analysis grid is constructed using the passenger objects included in the second orientation image. Then, the similarity between the first analysis grid and the second analysis grid is compared to obtain a similarity comparison result. Finally, the second analysis grid is adjusted according to the similarity comparison result to determine the distinguishing grid points in the second analysis grid.

[0117] After the passenger object recognition process is completed, the first analysis grid is constructed using the passenger objects included in the first orientation image and the second analysis grid is constructed using the passenger objects included in the second orientation image.

[0118] When constructing the first analysis grid and the second analysis grid, the features of the analysis series are first used for association. Of course, features such as special patterns (on clothes), clothing colors, clothing textures, etc. can also be used for association. The grid points obtained at this time are the reference points, and then the passenger objects that cannot be associated are connected to the reference points.

[0119] For the first analysis grid and the second analysis grid, the grids in the two analysis grids need to use the same shape with the same number of sides, generally using triangles. Then compare the similarity of the first analysis grid and the second analysis grid, and adjust the first analysis grid and the second analysis grid by zooming in, zooming out, and translating during the comparison, so that the first analysis grid and the second analysis grid overlap as much as possible.

[0120] If at this time there are grid points that cannot be overlapped in the first analysis grid or the second analysis grid, the passenger object corresponding to the grid point is used as a newly added passenger object.

[0121] In some examples, the associated feature objects belonging to the newly added passenger object are determined by:

[0122] S501, establishing a simplified model of a newly added passenger object, where the simplified model includes a head, a torso, and four limbs;

[0123] S502, establishing a suspected region based on the limbs of the simplified model;

[0124] S503, searching for a related feature object in the suspected area, and when a related feature object exists in the suspected area, binding the found related feature object to the newly added passenger object.

[0125] In step S501 to step S503, a simplified model of the newly added passenger object is established, the simplified model includes the head, torso and limbs, and then the suspected area is established based on the limbs of the simplified model, such as Figure 6 As shown, finally, the associated feature objects are found in the suspected area.

[0126] As mentioned in the foregoing content, for a newly added passenger object, the head can be identified relatively easily, and then the hands and legs associated with the head are searched based on the head, that is, the limbs of the simplified model mentioned in step S502.

[0127] Then the extremities of the simplified model were used as references to establish the suspected region.

[0128] The purpose of this method is to quickly determine the potential existence area of ​​the associated feature object.

[0129] The method of finding associated feature objects in the suspected area is as follows:

[0130] S601, dividing the suspected region by color to obtain a plurality of first suspected sub-regions, wherein the color of pixels in each first suspected sub-region tends to be consistent;

[0131] S602, screening the first suspicious sub-region once using the shape to obtain a second suspicious sub-region;

[0132] S603, picking up contour features in the second suspicious sub-region and establishing a feature surface based on the contour features;

[0133] S604, picking up charging features on the feature surface;

[0134] S605, performing voting calculation on the contour features and the charging features to obtain a calculation result;

[0135] S606: The type of the associated feature object is given according to the calculation result.

[0136] In step S601 to step S606, the suspected region is first divided by color to obtain a plurality of first suspected sub-regions. For the first suspected sub-regions, the color of the pixels in each first suspected sub-region is required to be consistent, that is, the first suspected sub-region is a monochrome region.

[0137] It should also be noted here that a portion of the suspected area that is equal to the set area needs to be incorporated into other suspected areas.

[0138] Then, the shape is used to screen the first suspicious sub-region to obtain the second suspicious sub-region, and then contour features are picked up in the second suspicious sub-region and a feature surface is established based on the contour features. Picking up contour features is to obtain contour features of the object included in the first suspicious sub-region, which is generally the outer contour shape of the first suspicious sub-region.

[0139] At the same time, the charging features are picked up on the feature surface. The charging features include handles, charging ports, trademarks, and patterns. Then the random forest model is used to vote on the contour features and charging features to obtain the calculation results. The random forest model is also a trained neural network, which is composed of multiple decision trees. The contour features and charging features are given to each decision tree, and each decision tree will also give a result. Finally, the results are counted and the result with the largest proportion is taken as the final result. In the end, there are two, which are electric vehicle rechargeable batteries and not electric vehicle rechargeable batteries.

[0140] At this point, the type of the associated feature object can be given based on the final result.

[0141] The embodiments of this specific implementation method are all preferred embodiments of the present application, and are not intended to limit the protection scope of the present application. Therefore, all equivalent changes made based on the structure, shape, and principle of the present application should be included in the protection scope of the present application.

Claims

1. An intelligent controller, characterized in that: include: Control unit (1); A first image acquisition unit (2) and a second image acquisition unit (3), wherein the first image acquisition unit (2) and the second image acquisition unit (3) are both electrically connected to the control unit (1); The first image acquisition unit (2) and the second image acquisition unit (3) are used to feed back image data to the control unit (1); The control unit (1) is used for obtaining a newly added passenger object and an associated characteristic object belonging to the newly added passenger object according to the obtained image data analysis, and issuing a control instruction according to the type of the associated characteristic object; Among them, control instructions include run, stop and warning reminders.

2. The intelligent controller according to claim 1, characterized in that: The control unit (1) obtains, based on the obtained image data, a newly added passenger object and associated feature objects belonging to the newly added passenger object, including: In response to an opening instruction of the elevator car, a first orientation image and a second orientation image are obtained, wherein both the first orientation image and the second orientation image point to a door opening area; identifying a passenger object included in the first orientation image and a passenger object included in the second orientation image; Associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain a newly added passenger object; Determine the associated feature objects that belong to the newly added passenger object.

3. The intelligent controller according to claim 2, characterized in that: Passenger identification objects include: Acquire a first reference image and a second reference image in a time series, wherein the generation time of the first reference image is earlier than the generation time of the second reference image; Comparing the first reference image with the second reference image, and determining a difference area on the first reference image or the second reference image; It is determined whether there is a passenger object in the distinguished area, and when the passenger object exists in the distinguished area, the passenger object is marked.

4. The intelligent controller according to claim 3, characterized in that: When determining whether there is a passenger object in the distinguishing area, a facial recognition method is used for determination, and the facial recognition method includes: determining the contours of the distinguishing regions and searching for facial features within the contours; Find facial details among facial features; Establish an analysis line based on facial detail features and use the pixel points on the analysis line to construct an analysis series; Analyze the analytical sequence and obtain the characteristics of the analytical sequence; It is determined whether there is a passenger object in the distinguished area based on the characteristics of the analysis sequence.

5. The intelligent controller according to claim 4, characterized in that: The characteristics of analyzing the analytical sequence and obtaining the analytical sequence include: Calculate the difference sequence or quadratic difference sequence of the analytical sequence; Determine a slope change region on the difference sequence or the quadratic difference sequence, wherein the number of the slope change regions is multiple; Calculate the distance between adjacent slope change areas and record it as the characteristic distance; According to the characteristic distance, the reference point of the analysis line is adjusted and the slope change area is calculated again to make the characteristic distance approach the characteristic distance model; When the distance between two slope change regions satisfies the characteristic distance model, other slope change regions between the two slope change regions are deleted.

6. The intelligent controller according to claim 2, characterized in that: Associating the passenger object included in the first orientation image with the passenger object included in the second orientation image to obtain a newly added passenger object includes: constructing a first analysis grid using passenger objects included in the first orientation image and constructing a second analysis grid using passenger objects included in the second orientation image; Comparing the similarity between the first analysis grid and the second analysis grid to obtain a similarity comparison result; adjusting the second analysis grid according to the similarity comparison result, and determining the distinguishing grid points existing in the second analysis grid; The passenger object corresponding to the distinguished grid point is used as the newly added passenger object.

7. The intelligent controller according to claim 2, characterized in that: The associated feature objects determined to belong to the newly added passenger object include: Establish a simplified model of the newly added passenger object, which includes the head, torso and limbs; The suspected area was established based on the extremities of the simplified model; Search for associated feature objects in the suspected area. When an associated feature object exists in the suspected area, bind the found associated feature object to the newly added passenger object.

8. The intelligent controller according to claim 7, characterized in that: Finding related feature objects in the suspected area includes: Dividing the suspected region by color to obtain a plurality of first suspected sub-regions, wherein the color of pixels in each first suspected sub-region tends to be consistent; Use the shape to screen the first suspicious sub-region once to obtain the second suspicious sub-region; Pick up contour features in the second suspicious sub-region and establish a feature surface based on the contour features; Pick up charging features on the feature surface; Perform voting calculation on the contour features and the charging features to obtain calculation results; The type of the associated feature object is given according to the calculation result.

9. An elevator, characterized in that: Comprising an intelligent controller as claimed in any one of claims 1 to 8.

Images