JPH08127476A - Method of detecting number of waiting passenger for elevator - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a method for accurately detecting the number of passengers waiting at gateways for an elevator by using an image processing system. SOLUTION: A difference image which shows a difference in brightness between an input image obtained by photographing waiting passengers with a camera set up in position at gateways on floors and a reference image obtained by photographing no waiting passengers is found (S3), the input image is divided into areas where the passengers are photographed by using the input image and the difference image (S4), the degree of confirmation that heads are imaged in the areas is computed (S5, S6) and the areas where the degrees of confirmation are over a preset value are counted (S7). Picture elements shared by heads or bodies photographed in the areas divided by using geometric modelling are extracted (S8) whereby the areas where the bodies are imaged in the input image are removed (S9) and the picture elements shared by heads are computed (S10). Waiting passengers are detected from the number of the areas found in the processes of S7, S10.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting the number of waiting passengers in an elevator management control system, and more particularly, to a method for accurately detecting the number of waiting passengers to get on an elevator using an image processing system. The present invention relates to a method for detecting the number of waiting passengers of an elevator to be obtained.

[0002]

2. Description of the Related Art The rapid development of the hardware and software industries related to computer-related technology related to elevator group management control systems has enabled advanced group management control. The core of the group management control system is that when a user presses a call button at an entrance / exit to use an elevator, an elevator that is in an optimal state among the attached elevators is assigned to respond to the call. It is. To implement such an algorithm for allocating an optimal elevator, a number of factors are required, and the most central factor among them is to predict the traffic volume. In other words, it can be said that the accuracy of the prediction for the number of passengers getting on and off from the entrance / exit until the elevator arrives at the relevant layer where the call is generated determines the reliability of the allocation algorithm.

In order to predict the traffic volume, it is necessary to detect the number of passengers waiting at the entrance / exit for boarding the elevator at each floor of the building. The method of detecting the number of waiting passengers is roughly classified into the following two methods. One is based on the existing accounting method based on the collected information, and the other is based on a vision system using image processing technology.

FIG. 5 is a block diagram showing an apparatus for detecting the number of waiting passengers using a conventional method. In FIG. 5, the detection device includes a data generation unit 100 that generates data indicating the number of passengers getting on and off at each layer when a user uses the elevator, learning necessity and learning degree of the data generated from the data generation unit 100. According to the learning reference value generation unit 101 that generates a reference value for determining, and the reference value for the learning necessity and the degree of learning that is generated from the learning reference value generation unit 101, the data generation unit 100 is used to get on and off. A data learning unit 102 that modifies passenger information, a traffic flow database unit 103 that stores data on the number of passengers who have been modified by the data learning unit 102, and a traffic flow database when a call occurs from an arbitrary entrance / exit Using the information stored in the section 103, the elevators of each layer until the elevator arrives in response to the call at the relevant entrance / exit And a traffic flow prediction unit 104 for calculating a traffic flow indicating the predictive distribution of the number of passengers getting on or off the motor.

In the conventional apparatus for detecting the number of passengers on standby using the total accounting method, first, when a user uses an elevator, the data generator 100 calculates the number of passengers who got on and off at each floor of the building. The generated data is output to the data learning unit 102. The data learning unit 102 reads the reference value information provided from the learning reference value table of the learning reference value generation unit 101 to determine the necessity of learning and the degree of learning, and according to the determined necessity of learning and the degree of learning,
Information about the number of passengers getting on and off generated from the data generating unit 100 is modified, and the number of passengers getting on and off for each layer is output to the traffic flow database unit 103. When a call is generated from an arbitrary entrance in a state where such traffic flow data is stored, the traffic flow prediction unit 104 uses the information on the number of getting on and off the passengers stored in the traffic flow database unit 103, and The distribution of the number of passengers getting on and off the elevator at each layer until the elevator arrives in response to the call is calculated to predict the traffic flow.

[0006] When a user presses a push button installed at a doorway to use an elevator, in order to allocate an optimal elevator to the call of the doorway, it is necessary to wait until the elevator arrives at the corresponding floor. The traffic flow indicating the distribution of the number of passengers who actually get on and off at each layer during the period should be predicted in advance. However, according to the conventional method for detecting the number of waiting passengers using the legal total accounting method described above, since the prediction of the traffic flow is performed using the total information based on the information collected previously, the actual instantaneous Traffic changes cannot be reflected. Further, there is a problem that the accuracy of the predicted data is inferior and the reliability is reduced.

Therefore, in order to improve the accuracy and the prediction of the actual traffic flow, a method of detecting the number of waiting passengers using a vision system based on image processing is used as shown below. FIG. 6 is a block diagram showing a configuration of a detection device using a conventional method of detecting the number of waiting passengers using a vision system. In FIG. 6, the detection device is provided at the entrance of each layer and includes a camera group 200 composed of a plurality of cameras for photographing a waiting passenger, and converts a signal captured by the camera group 200 into video information and transmits the video information to the communication bus 205. A camera control unit 201, a program ROM 202 storing a program for calculating the number of passengers getting on and off using the video information transmitted from the camera control unit 201, and a program R
According to the program of the OM 202, the video information is arithmetically processed to determine the number of passengers who got on and off, and the central processing unit 203 for transmitting data indicating the obtained number of passengers getting on and off to the communication bus 205, and the central processing unit 203 The data RAM 204 stores data indicating the number of passengers and information on the position of the camera group 200.

FIG. 7 is a flowchart showing a process of detecting the number of waiting passengers performed by the central processing unit 203 of FIG. In the figure, when video information captured by the camera group 200 is input to the central processing unit via the communication bus 205, the camera group 200 stored in the data RAM 204
Information on the position, i.e., an input image analysis process of analyzing the image information using the height and distance of the camera,
In the analyzed image information, a threshold value comparison process of deleting pixels whose brightness is equal to or greater than a predetermined threshold value, and information about brightness based on the brightness difference of each pixel of the image remaining in the threshold value comparison process, A distribution mode test (Partition Mo) that converts information about the distance between captured objects and divides or integrates the image area using this distance information
de Test: PMT) division process, head confidence calculation process for calculating how similar the shape of the subject photographed in the integrated or divided region is to the shape of a human head, and the calculated head The method includes a process of calculating the number of regions that counts a region having a certain value or more in the certainty factor, and a process of detecting the number of passengers that determines the obtained number of regions as the number of passengers.

In the method of detecting the number of waiting passengers using the conventional vision system having such a process, first, when the user presses a call button installed at the entrance to use the elevator, the entrance of each layer is entered. A camera group 200 composed of cameras installed in the respective sections outputs signals to the camera control section 201, which capture images of the passengers waiting at the entrances of the respective layers. The camera control unit 201 converts a signal captured by a camera of each layer into a video signal, and outputs the video signal to the central processing unit 203 via the communication bus 205. The central processing unit 203 is the communication bus 2
The video signal of the camera control unit 201 input via 05 is analyzed according to the flowchart of FIG. 7, and the number of passengers currently waiting in each layer is detected.

That is, when a video signal is input to the central processing unit 203 via the communication bus 205, the central processing unit 20
Reference numeral 3 reads information on the position of each camera stored in the data RAM 204, that is, information on the height and shooting distance of a camera installed at the entrance of each layer, and analyzes the video signal based on the information. Then, the brightness of each pixel of the analyzed video signal is compared with a critical value, and the pixels having the brightness equal to or higher than the critical value, that is, the pixels other than the passenger's head are photographed are deleted from the target. When the pixels having the threshold value or more are deleted, the information on the brightness based on the brightness difference is converted into the information on the distance between the photographed subjects by the distribution mode test division process, and the area is converted using the converted distance information. Is integrated or divided.

More specifically, first, the distance information M (A, B) is obtained by the following equation: M (A, B) = | I (A) -I (B) | Here, I (A) is the brightness of the area A of the video signal, and I (B) is the brightness of the area B adjacent to the area A of the video signal. Distance information M (A, B) obtained from the above equation
Is small, that is, when the difference in lightness is small, it is determined that the distance between the subjects photographed in the two areas is short, the two areas (A, B) are integrated, and the distance information M (A, When the value of B) is large, that is, when the brightness difference is large, it is determined that the distance between the subjects photographed in the two regions is long, and the two regions (A, B) are divided. When this process is repeated, the information about the brightness of the video signal is converted into information about the distance, and the video is divided into a plurality of regions. The distinguished region is shifted to the next head confidence calculation process. .

In the process of calculating the head certainty factor, the subject photographed in each of the obtained areas is determined using the difference in brightness between the hair part and the skin (face) part and the shape of the round head. A head certainty factor indicating how close the head is is calculated. Next, in the calculation process of the number of regions, only the region having a predetermined value or more among the head reliability of each region obtained in the calculation process of the head reliability is counted. In the process of detecting the number of passengers, the number of regions counted in the process of calculating the number of regions is determined as the number of passengers. That is, for example, when there are two regions where the head certainty is equal to or more than the predetermined value, it is determined that the number of passengers is two.

However, in such a conventional apparatus for detecting the number of passengers on standby using a vision system, it is desirable that a camera be installed at the upper center of the entrance and take a picture of the entrance and exit. Due to the low height of the passenger, it is not possible to photograph the waiting passenger at the upper center of the entrance, and therefore, the camera is installed on the upper left or right side of the entrance and the passenger waiting at the entrance There are cases where only the side of the camera is shot. In such a case, there is a possibility that a part of the head of the waiting passenger is superimposed and photographed, and when this superimposition occurs, an error occurs in the detected number of passengers and the accurate number of passengers cannot be detected. There was a problem.

Further, since the number of passengers is detected based only on the information regarding the brightness of the video signal, the accuracy of the detection is affected by the change in the reference value of the brightness, and the probability of occurrence of errors increases. there were. Therefore, the present invention obtains a difference image which is a difference between an input image photographed by a camera at a predetermined entrance and a reference image photographed beforehand when there is no passenger at the entrance, and calculates the difference image between the input image and the difference image. It is an object of the present invention to provide a method for detecting the number of waiting passengers of an elevator, which detects the number of waiting passengers by using the above as a calculation target.

Further, the present invention calculates the number of pixels occupied by an area having a high degree of head certainty in an input video, and performs geometric modeling based on the calculated number of pixels, the position of a camera, and a human body structure. It is an object of the present invention to provide a method for detecting the number of waiting passengers of an elevator, which can accurately detect the number of waiting passengers based on the number of pixels occupied by the head of a subject which can be photographed in the area calculated using the number of waiting passengers.

[0016]

According to the first aspect of the present invention, there is provided a method for detecting the number of waiting passengers of an elevator using a video processing system, wherein the number of passengers is determined by a camera installed at a predetermined entrance. An input image, based on a reference image captured when there is no waiting passenger at the entrance / exit in advance, a video comparing step of obtaining a difference image indicating a difference between the input image and the reference image; and A distribution mode test division process of obtaining information on a distance of a subject photographed at an adjacent pixel of the input image from information on a difference in brightness of the difference image, and dividing and integrating an area of the input image with the distance information; A head confidence that calculates a certainty for a predetermined body part with respect to the area divided and integrated in the distribution mode test distribution process, and calculates a head certainty based on the certainty And operation process, the head confidence counts the area is a predetermined value or more, characterized in that the number is the count area comprising an arithmetic process number of areas determines the waiting passengers.

According to a second aspect of the present invention, in the first aspect of the present invention, the image comparing step includes: an input image photographed when a standby passenger is present at a predetermined entrance; and no standby passenger at the entrance. It is characterized in that a difference in brightness at the same position as a reference image that is sometimes taken is obtained as an absolute value. In the invention according to claim 3, in the invention according to claim 1 or 2, the distribution mode test division step includes: a first step of calculating, as an absolute value, a difference in brightness between two adjacent pixels of the input image; A second step of obtaining, as an absolute value, a difference in brightness of the difference image at the same position as the two adjacent pixels, and a second step of obtaining a value indicating the distance information from the values obtained in the first and second steps. Three steps, a fourth step of comparing the value indicating the distance information with a predetermined value, and regarding the comparison result, when the value indicated by the distance information is smaller than the predetermined value, the two adjacent pixels are placed in the same area. And integrating the adjacent two pixels into different areas when the distance information indicates greater than the predetermined value.

In the invention according to claim 4, in the invention according to claim 3, in the third step, the two adjacent pixels of the input image are P1 and P2, and the brightness of the pixel is I (P1 ), I (P2), and the pixel P of the input image
1, the brightness of the pixels of the difference image at the same position as P2 is represented by DI (P1), DI (P2), and the difference between the brightness of the pixels of the input image and the brightness of the pixels of the difference image The value indicating the distance information is represented by the following equation: α | I (P1) −I (P2) | + (1−α) | DI (P
1) -DI (P2) |

[0019] In the invention according to claim 5 is the invention according to claims 1 to 4, the operation process of the head confidence in split and integration resulting in the distribution mode test division process area
For, the position and determining a hair confidence, below the region adjacent to the region with a hair confidence of該髪region Nitsu
And a step of obtaining a face certainty factor, and a region having the head certainty factor and a region having the hair certainty factor and a region having the face certainty factor, and the head certainty factor of the region having the head certainty factor is the hair certainty factor. It is characterized by comprising a step of summing up the face certainty factor and a step of obtaining a body certainty factor for a region excluding a region having the head certainty factor.

According to a sixth aspect of the present invention, in the invention according to the fifth aspect, an area having a high head certainty obtained in the head reliability calculation process is defined as a head, and an area located below the area is defined as a body. Assuming that the method includes a fuzzy mitigation process for increasing the head certainty factor. The invention according to claim 7 is a method for detecting the number of waiting passengers of an elevator using an image processing system, comprising: an input image photographed by a camera installed at a predetermined entrance and exit; Based on the reference video taken when not, based on the video comparison process of obtaining a difference video indicating the difference between the input video and the reference video, and information on the difference in brightness between the input video and the difference video, Find information about the distance of the subject photographed in the adjacent pixels of the input video,
Based on the certainty factor, a distribution mode test dividing process for dividing and integrating the region of the input image based on the distance information, and a certainty factor for a predetermined body part in the divided and integrated region in the distribution mode test distribution process are obtained. The number of pixels occupied by the head of the subject and the pixels occupied by the body that can be photographed in the area by the process of calculating the head certainty factor and the geometric modeling based on the position of the camera and the human body structure. And a geometrical feature extraction process for calculating the number of pixels, and the head certainty factor obtained in the head certainty factor calculation process based on the number of pixels occupied by the body obtained in the geometrical feature extraction process. Is a body removal process for removing a region equal to or less than a predetermined value, the number of pixels included in the region left unremoved in the body removal process, and the pixels occupied by the head calculated in the geometric feature extraction process Number and A calculation process of the number of pixels for detecting the number of passengers in comparison, characterized in that it comprises a.

According to an eighth aspect of the present invention, in the invention according to the seventh aspect, in the step of calculating the number of pixels, the area remaining without being removed in the body removing step is represented by i, and the number of remaining areas is represented by i. N, the number of pixels in the remaining area is P ′ (i),
The number of pixels occupied by the head calculated in the process of extracting the geometric features is defined as P '' (i), and the number of standby passengers N '' is expressed by the following equation:

[0022]

[Equation 2]

It is characterized by the following. Claim 9
In the invention according to claim 1, there is provided a method for detecting the number of waiting passengers of an elevator using an image processing system, wherein an input image taken by a camera installed at a predetermined entrance and exit and an image taken when there are no waiting passengers at the entrance and exit in advance. The image comparison process for obtaining a difference image showing the difference between the input image and the reference image based on the reference image that has been input, and information on the difference in brightness between the input image and the difference image based on the information about the difference in brightness between the input image and the difference image. A distribution mode test division process of obtaining information about the distance of an object photographed in adjacent pixels, dividing and integrating an area of an input image based on the distance information, and geometric modeling based on the position of the camera and the human body structure. The geometric feature extraction process for calculating the number of pixels occupied by the head and the number of pixels occupied by the body of the subject that can be photographed in the area, and the distribution mode. Calculating a certainty factor for a predetermined body part with respect to the region divided and integrated in the test distribution process, and calculating a head certainty factor based on the certainty factor; The head confidence obtained in the head confidence calculation process is calculated based on the number of pixels occupied by the body obtained in the step of calculating the number of regions, and the number of pixels occupied by the body obtained in the step of extracting the geometric feature. A body removal process of removing an area that is equal to or less than a predetermined value, the number of pixels included in a region remaining without being removed in the body removal process, and the number of pixels occupied by the head calculated in the extraction process of the geometric feature. And a step of extracting the number of passengers for extracting the final number of passengers from the result obtained in the pixel number calculation step and the result obtained in the area number calculation step. It is characterized by comprising.

In the invention according to claim 10, in the invention according to claim 9, in the process of detecting the number of passengers, the result obtained in the process of calculating the number of areas is N ′, and is obtained in the process of calculating the number of pixels. It is characterized in that the final number of passengers is obtained by βN ′ + (1−β) N ″, where N ″ is the result and β is a predetermined weight value.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings. FIG. 1 is a schematic flowchart for explaining a method of detecting the number of waiting passengers of an elevator using the video processing system according to the embodiment of the present invention, and FIG. 2 is a method of detecting the number of waiting passengers of the elevator of FIG. FIG. 2 is a block diagram illustrating a configuration of a detection device that utilizes the above.

In FIG. 2, the detection device selects a camera group 300 which is installed at a predetermined position at the entrance of each layer and includes a plurality of cameras for photographing the number of waiting passengers, and a signal photographed by the camera group 300. Output camera selection unit 3
01 and the signal selected by the camera selection unit 301 are converted into video signals to detect the number of waiting passengers, and the video signals and the detection results are transmitted to the cable television monitor, the result display unit, and the group management control system. And a vision system 302 including a central processing unit.

First, a first embodiment according to the first to sixth aspects of the present invention will be described. In FIG. 1, the method for detecting the number of waiting passengers according to the first embodiment includes an input image capturing step S1 for capturing an input image of a passenger waiting at an entrance / exit of each layer; A reference image taken in advance when there is no reference image, a reference image importing step S2, a video comparing step S3 for obtaining a difference image from a difference between the input image and the reference image, and a difference image and an input obtained in the video comparing step S3. A distribution mode test division step S4 of converting the information on the brightness of the input image into information on the distance of the photographed subject using the image, and dividing or integrating the area of the input image by the distance information;
For each region obtained in the distribution mode test division process S4,
A head confidence factor calculation process S5 for calculating the confidence factor of the body, face, head, and body parts of the hair based on the size of each region, the ratio of length × width, the difference in brightness and the like, and calculating the head confidence factor. A fuzzy relaxation process S6 for increasing the head confidence obtained by the head confidence calculation process S5 by using the correlation between the head and the body, and a head confidence increased by the fuzzy relaxation process S6. The number of regions is calculated by calculating the number of regions S7.

The operation and effect of the first embodiment having such a process will be described. First, when the user presses a button at the entrance to use the elevator, a camera group 300 including a plurality of cameras installed at the entrance of each layer transmits a signal photographing a passenger waiting at the entrance of each layer. Output to the camera selection unit 301. The camera selection unit 301 selects a video signal obtained by the camera group 300 actually photographing a waiting passenger, and transmits the video signal to a vision system 302 including a central processing unit via a communication bus. The vision system 302 takes in, via a communication bus, an input image of a standby passenger actually taken, and takes in, via a camera selection unit 301, a reference image taken in advance when there is no passenger at the corresponding entrance. Here, the input video image of the number of waiting passengers and the reference video image are video signals captured at the same place. When the reference image and the input image are obtained in this manner, the vision system 302 detects the number of waiting passengers by a method according to the schematic flowchart of FIG. 1 and the flowchart of FIG.

A detailed description will be given below with reference to examples. First,
As shown in FIG. 3, the vision system 302 determines a difference image between the input image and the reference image as an absolute value of a brightness difference between pixels at the same position, as shown in FIG. In this way, when the difference image is calculated, the absolute value of the brightness difference is used and the person is divided from the background based on the magnitude of the absolute value. Therefore, whether the background portion of the input image is darker or brighter than before. Information about can be ignored. This means that people are photographed darker than buildings,
In the case of classifying people based only on the difference between dark and bright, when the background of the input video becomes darker than the background of the reference video, the background portion of the input video is erroneously detected as a human. However, if a person is detected based on the magnitude of the absolute value of the brightness difference, there is no mistake. Therefore, in the present invention, the magnitude of the absolute value of the brightness difference is distinguished, and particularly, difference information having a large absolute value, that is, darker, is used mainly.

Next, when the difference image is obtained, in the distribution mode test division step S4, the input image region is divided or integrated using the input image and the difference information. That is, two adjacent pixels P1 of the input image as shown in the example of FIG.
The brightness difference A'of P2 is obtained as an absolute value. That is,
The difference A ′ in brightness between two pixels P1 and P2 adjacent to each other in the input image is obtained by | I (P1) −I (P2) |. Where I (P1) is the brightness of pixel P1 in the input video,
I (P2) is the brightness of the pixel P2.

In this way, when the brightness difference A ′ between the pixels of the input image is obtained, the brightness difference B ′ between two adjacent pixels in the difference image is obtained as an absolute value.
That is, the brightness difference DI (P1) between the pixel P1 of the input image and the pixel P1 of the reference image, and the brightness difference DI between the pixel P2 of the input image and the pixel P2 between the reference images, which are obtained as the difference images. The difference B ′ from (P2) is obtained as an absolute value. In other words, in the brightness difference between the input image and the reference image at the pixel at the same location, the brightness difference DI
(P1), the difference B 'between DI (P2) and | DI (P1)-
DI (P2) |

When the brightness difference A 'between the pixels in the input image and the brightness difference B' between the pixels in the difference image are obtained, the two pixels are photographed by using these. Value M indicating information on distance between subjects
(P1, P2) is obtained. Here, the distance information M (P1,
P2) is given by M (P1, P2) = α | I (P1) −I (P2) | +
(1−α) | DI (P1) −DI (P2) | Here, α is a coefficient for adjusting the difference in brightness between pixels in the input image and the importance of the difference in brightness between pixels in the difference image.

As described above, the distance information M (P1, P
When 2) is obtained, the distance information is compared with predetermined distance information, and the two pixel regions are divided or integrated. That is, when the value of M (P1, P2) is larger than the predetermined distance information, that is, when the difference in brightness between the two pixels P1 and P2 of the input image is large, the distance between the objects captured by the two pixels P1 and P2 is large. Are judged to be distant, and they are recognized as different areas and divided. On the contrary, if the value of M (P1, P2) is smaller than the predetermined distance information, that is, if the difference in brightness between the two pixels P1 and P2 is small, the distance between the objects photographed in the two pixels P1 and P2 is close. Judged, recognized as the same area, and integrated. This process is applied to all the pixels of the input image. As a result, the first to sixth regions A1 to A6 shown in the lower part of FIG. 3 are obtained, and the process proceeds to the next head certainty calculating process S5.

In the head confidence calculation process S5, FIG.
As shown in (5), the hair confidence A, the face confidence B, the head confidence C, and the body confidence D are obtained for all the areas A1 to A6 obtained in the distribution mode test division step S4. here,
This will be specifically described with reference to FIGS. In the head confidence calculation step S5, the distribution mode test division step S4
First, the hair credibility A is determined using the width of each area, the ratio of height to width, the difference between light and dark, and the like with respect to the first to sixth areas A1 to A6 obtained in the above. For example, it is assumed that the first and second areas A1 and A2 in FIG. Then, the face certainty factor B of the area adjacent to and below the first and second areas A1 and A2 is calculated. It is assumed that the third and fourth regions A3 and A4 in FIG. After obtaining the hair credibility A and the face credibility B in this manner, an area having a high hair credibility A and a face credibility B is defined as a high head credibility area, and the hair credibility A and the face credibility B are determined. Summed head confidence C
Ask for. That is, the first and third regions A1, A3 and the second,
A head certainty factor C is obtained from each of the fourth regions A2 and A4.
Next, when the head certainty C is obtained, the body certainty D is obtained. For example, the fifth and sixth regions A5 and A6 that are adjacent to and below the third and fourth regions A3 and A4 have the body certainty D
Is high. When the head confidence factor calculation process S5 is completed for all the regions in this way, they are provided to a fuzzy mitigation process S6.

In the fuzzy mitigation step S6, the first to sixth areas A1 to A6 obtained in the head confidence calculation step S5.
Increase the head confidence C of. That is, the first and third areas A1 and A3 and the second and fourth areas A2 and A4 having high head certainty C obtained in the head certainty calculating step S5 are set as heads, and are located adjacent to and below the areas. Fifth and sixth areas A5
After assuming that A6 is a body, a certainty factor D ′ that the fifth and sixth regions A5 and A6 are bodies is obtained using the correlation between the head and the body. When the body certainty factor D ′ thus obtained is high, the first and third regions A1 and A3 and the second and fourth regions A2 and A4 are
Is again calculated and increased. Next, when the fuzzy mitigation process S6 is completed for all the regions, the process proceeds to the body removing process S9 and the region number calculating process S7.

In the step S7 of calculating the number of regions, the regions in which the degree of head certainty C obtained in the fuzzy mitigation process S6 is equal to or larger than a predetermined value are counted and set as the number of regions N '. Then, the number of passengers is detected using the obtained number of areas N '. For example, when the number of areas N ′ is 2, the number of passengers is detected as 2 persons. When the number of passengers is detected in this way, the central processing unit of the vision system 302 displays the result in the result display unit,
Transmission to cable television and group management control system.

Next, a second embodiment according to the present invention will be described. Second embodiment and first
The process different from the first embodiment is that steps S8 to S10 are added instead of the step S7 after the step S6 of the first embodiment. Therefore, S1 of the second embodiment
Since the process of S6 is the same as that of the first embodiment, the description is omitted. The hardware configuration is the same as in FIG.

In FIG. 1, after the fuzzy mitigation process S6, a subject that can be photographed in each area obtained in the distribution mode test division process S4 by geometric modeling based on the position, distance, size of a person, etc. of the camera. The geometric feature extraction step S8 for calculating the number of pixels occupied by the head and the number of pixels occupied by the body, and the information obtained in the fuzzy mitigation step S6 and the information obtained from the geometric feature extraction step. The body removal step S9 for removing the portion of the input video where the body was photographed, and the head number obtained in the step S8 for extracting the number of pixels and the geometrical features of the region left unremoved in the body removal step S9. And calculating the number of pixels to be compared with the number of pixels occupied by.

The operation and effect of the second embodiment is, in addition to the operation and effect of the first embodiment, in order to make the detection of the number of waiting passengers more accurate, in the geometric feature extraction step S8, The geometrical modeling based on the size and the position of the camera is used to calculate the number of pixels P ″ occupied by the head of the subject and the number of pixels occupied by the body of the subject that can be photographed in the first to sixth areas A1 to A6. The removal process S9 is provided.

In the body removal step S9, based on the number of pixels occupied by the body provided in the geometric feature extraction step S8, the head certainty C obtained in the fuzzy relaxation step S6 is equal to or larger than a predetermined value. 1, third regions A1, A3 and second,
All regions except the fourth regions A2 and A4, that is, the fifth and sixth regions A5 and A6 corresponding to the body are removed. In the process S10 of calculating the number of pixels, the regions that remain without being removed in the body removing process S9 belong to the first and third regions A1 and A3 and the second and fourth regions A2 and A4 corresponding to the head. The number of pixels P ′ is calculated. Then, a result value N ″ is obtained as shown in Expression 3 which is obtained by summing the values obtained by dividing the number of pixels P ′ and the number of head pixels P ″ obtained in the geometric feature extraction step S8 for each area. .

[0041]

(Equation 3)

Here, N is the number of regions remaining without being removed. By doing so, for example, when the number of pixels of the head appearing in the image is 70 due to the overlapping of the head, this is conventionally recognized as one person, but using the result value N ″, the extraction of geometric features is performed. By calculating the result value N ″ as 1.4 persons by using the number of pixels of the head portion by the geometrical modeling calculated in step S8, for example, the number of pixels of 50, the number of detected passengers is further increased. Be accurate.

Next, a third embodiment according to the invention described in claims 9 and 10 will be described. The process in which the third embodiment is different from the first and second embodiments is that the process of S7 of the first embodiment and the process of S10 of the second embodiment are performed after S7.
This is the point where 11 processes have been added. Therefore, steps S1 to S10 of the third embodiment are the same as those of the first and second embodiments, and a description thereof will be omitted. The hardware configuration is shown in Figure 2.
Is the same as

In FIG. 1, in the third embodiment, S1.
After the step S10, a passenger number detection step S11 is performed to detect the number of passengers using the comparison result obtained in the pixel number calculation step S10 and the area number obtained in the area number calculation step S7. The operation and effect of the third embodiment are the same as the operation and effect of the first and second embodiments. In addition, in order to detect the number of waiting passengers more accurately, in the passenger number detection step S11, the number of areas is calculated S7. The number N 'of regions having a high head certainty C obtained in the above and the result value N''obtained in the process S10 of calculating the number of pixels are applied to the following equation to detect the final number of passengers.

(Number of passengers) = βN ′ + (1−β) N ”where β is a weight value that is intermediate between the heights of the humans. If a more accurate number of passengers is detected in this way, the vision becomes The central processing unit of the system 302 transmits the result to the result display unit, the cable television, and the group management control system.

Note that the vision system 302 of FIG. 1 provides a user interface so that the used predetermined values (α value, β value, camera constant, etc.) can be adjusted.

[0047]

As described above, according to the first aspect of the present invention, the head reliability is obtained by using the difference image due to the difference between the input image and the reference image, and the number of waiting passengers can be more accurately detected. There is. According to the inventions according to the second to sixth aspects, the function of the invention according to the first aspect can be efficiently achieved.

According to the seventh aspect of the present invention, when the number of pixels in an area where the head confidence is equal to or greater than a predetermined value is divided by the number of pixels extracted from the geometric modeling process, and the number of waiting passengers appears in a superimposed manner. , The number of passengers can be accurately detected. According to the invention of claim 8, the action of the invention of claim 7 can be efficiently exhibited.

According to the ninth aspect of the present invention, the final number of passengers can be detected more accurately by summing the value obtained from the area number calculation process and the value obtained from the pixel number calculation.
According to the invention of claim 10, the action of the invention of claim 9 can be efficiently exhibited.

[Brief description of drawings]

FIG. 1 is a schematic flowchart illustrating an operation of a method for detecting the number of waiting passengers of an elevator using a vision system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a configuration of a detection device using the method for detecting the number of waiting passengers according to the embodiment.

FIG. 3 is a diagram schematically illustrating a distribution mode test division process.

FIG. 4 is a flowchart showing the operation of FIG. 1 in detail.

FIG. 5 is a block diagram showing a configuration of an apparatus for detecting the number of waiting passengers of an elevator using a conventional accounting method.

FIG. 6 is a block diagram showing a configuration of an apparatus for detecting the number of waiting passengers of an elevator using a conventional vision system.

7 is a flowchart showing a process of extracting the number of waiting passengers in the central processing unit of FIG.

[Explanation of symbols]

 300 camera group 301 camera selection unit 302 vision system

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁶ Identification code Internal reference number FI technical display location G06M 11/00 D

Claims (10)

[Claims] 1. A method of detecting the number of waiting passengers in an elevator using an image processing system, comprising: an input image taken by a camera installed at a predetermined entrance and exit; and an image taken when there are no waiting passengers at the entrance and exit in advance. An image comparison process for obtaining a difference image showing a difference between the input image and the reference image based on the reference image, and information on a difference in brightness between the input image and the difference image, Information about the distance of the object photographed in adjacent pixels is obtained, and the distribution mode test division process of dividing and integrating the region of the input image based on the distance information and the region divided and integrated in the distribution mode test distribution process are determined. Of the certainty factor with respect to the body part of the human body, and the calculation process of the certainty factor based on the certainty factor; Counting the number of areas and calculating the number of areas as the number of waiting passengers. The method for detecting the number of waiting passengers of an elevator, comprising: 2. The image comparing step comprises the steps of: determining the brightness at the same position between an input image captured when a standby passenger is at a predetermined entrance and a reference image captured when no standby passenger is at the entrance. 2. The method according to claim 1, wherein the difference is obtained as an absolute value. 3. The distribution mode test division step includes: a first step of determining a difference in brightness between two adjacent pixels of the input image as an absolute value; and the difference image at the same position as the two adjacent pixels. A second step of obtaining a difference in brightness as an absolute value, a third step of obtaining a value indicating the distance information from the values obtained in the first step and the second step, and a value indicating the distance information being a predetermined value. And a fourth step of comparing the two pixels adjacent to each other into the same area when the value indicated by the distance information is smaller than the predetermined value, and indicating that the distance information indicates that the value is smaller than the predetermined value. 3. The method according to claim 1, further comprising: a fifth step of dividing the two adjacent pixels into different areas when the number is large. 4. In the third step, the two adjacent pixels of the input image are set to P1 and P2, and the brightness of the pixels is set to I1.
(P1), I (P2), and the pixel P of the input image
1, the brightness of the pixels of the difference image at the same position as P2 is represented by DI (P1), DI (P2), and the difference between the brightness of the pixels of the input image and the brightness of the pixels of the difference image The value indicating the distance information is represented by the following equation: α | I (P1) −I (P2) | + (1−α) | DI (P
The method for detecting the number of waiting passengers of an elevator according to claim 3, wherein the method is obtained by: 1) -DI (P2) | 5. The head confidence calculation step includes: obtaining a hair confidence for the divided and integrated region obtained in the distribution mode test division; and adjoining the region having the hair confidence. Obtaining a face certainty factor for an area located below the area; and setting the area having the hair certainty factor and the area having the face certainty factor as a region having the head certainty factor, and the head certainty of the region having the head certainty factor. 2. The method according to claim 1, further comprising: obtaining a degree of hair by adding the hair certainty degree and the face certainty degree; and obtaining a body certainty degree in an area excluding the area having the head certainty degree. 3.
The method for detecting the number of waiting passengers of an elevator according to any one of the above-mentioned items. 6. A fuzzy mitigation process for increasing the head certainty, assuming that a region having a high head certainty determined in the calculation process of the head certainty is a head and a region located below the region is a body. 6. The method for detecting the number of waiting passengers of an elevator according to claim 5, comprising: 7. A method for detecting the number of waiting passengers of an elevator using an image processing system, comprising: an input image captured by a camera installed at a predetermined entrance and exit; A video comparing step of obtaining a difference video indicating a difference between the input video and the reference video, based on the reference video, and information on a difference in brightness between the input video and the difference video. A distribution mode test division step of obtaining information relating to a distance of a subject photographed in an adjacent pixel, and dividing and integrating an area of an input image based on the distance information; and a predetermined division and integration area in the distribution mode test distribution step. Calculating a head certainty factor based on the certainty factor for the body part, and calculating a head certainty factor based on the certainty factor; and a position of the camera and a human body structure. A geometric feature extraction process for calculating the number of pixels occupied by the head of the subject and the number of pixels occupied by the body that can be photographed in the region, and A body removal step of removing a region in which the head confidence obtained in the head confidence calculation step is equal to or less than a predetermined value based on the number of pixels occupied by the body, and remaining without being removed in the body removal. And calculating the number of pixels for detecting the number of passengers by comparing the number of pixels included in the area with the number of pixels occupied by the head calculated in the step of extracting the geometric feature. A method for detecting the number of waiting passengers of an elevator. 8. In the process of calculating the number of pixels, the region remaining without being removed in the body removing process is represented by i, the number of remaining regions is N, and the number of pixels of the remaining regions is P ′ ( i), the number of pixels occupied by the head calculated in the geometric feature extraction process is P ″ (i), and the number N ″ of waiting passengers is expressed by the following equation: The method for detecting the number of waiting passengers of an elevator according to claim 7, wherein 9. A method of detecting the number of waiting passengers in an elevator using an image processing system, comprising: an input image taken by a camera installed at a predetermined entrance and exit; and an image taken when there are no waiting passengers at the entrance and exit beforehand. An image comparison process of obtaining a difference image showing a difference between the input image and the reference image based on the reference image, and information on a difference in brightness between the input image and the difference image, A distribution mode test division process of obtaining information on the distance of the object photographed in adjacent pixels, dividing and integrating the area of the input image based on the distance information, and geometric modeling based on the position of the camera and the human body structure. A geometric feature extraction process for calculating the number of pixels occupied by the head of the subject and the number of pixels occupied by the body that can be photographed in the region according to Calculating a head certainty for a predetermined body part with respect to the region divided and integrated in the test distribution process, and calculating a head certainty based on the certainty; The head confidence obtained in the head confidence calculation based on the number of pixels occupied by the body obtained in the geometric feature extraction process, A body removal process of removing an area that is equal to or less than a predetermined value, a number of pixels included in a region that is not removed in the body removal process, and a number of pixels occupied by the head calculated in the extraction process of the geometric feature. And a step of extracting the number of passengers for extracting a final number of passengers from a result obtained in the pixel number calculating step and a result obtained in the area number calculating step. Elevator features characterized by comprising Detection method. 10. In the step of detecting the number of passengers, the result obtained in the process of calculating the number of areas is N ′, the result obtained in the process of calculating the number of pixels is N ″, and a predetermined weight is β. As
The method according to claim 9, wherein the final number of passengers is obtained by βN ′ + (1−β) N ″.