EP0992949A2 - People counting system by pressure images - Google Patents

People counting system by pressure images Download PDF

Info

Publication number
EP0992949A2
EP0992949A2 EP99500166A EP99500166A EP0992949A2 EP 0992949 A2 EP0992949 A2 EP 0992949A2 EP 99500166 A EP99500166 A EP 99500166A EP 99500166 A EP99500166 A EP 99500166A EP 0992949 A2 EP0992949 A2 EP 0992949A2
Authority
EP
European Patent Office
Prior art keywords
images
people
pressure
footsteps
previous
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Withdrawn
Application number
EP99500166A
Other languages
German (de)
French (fr)
Other versions
EP0992949A3 (en
Inventor
Juan Ignacio Lopez Mudarra
Ramon Garcia Bayo
Carlos Palomo Gomez
Francisco Leon Sanchez
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Eliop Seinalia Sl
Original Assignee
ELIOP SA
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by ELIOP SA filed Critical ELIOP SA
Publication of EP0992949A2 publication Critical patent/EP0992949A2/en
Publication of EP0992949A3 publication Critical patent/EP0992949A3/en
Withdrawn legal-status Critical Current

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06MCOUNTING MECHANISMS; COUNTING OF OBJECTS NOT OTHERWISE PROVIDED FOR
    • G06M1/00Design features of general application
    • G06M1/08Design features of general application for actuating the drive
    • G06M1/10Design features of general application for actuating the drive by electric or magnetic means
    • G06M1/108Design features of general application for actuating the drive by electric or magnetic means by electronic means
    • GPHYSICS
    • G07CHECKING-DEVICES
    • G07CTIME OR ATTENDANCE REGISTERS; REGISTERING OR INDICATING THE WORKING OF MACHINES; GENERATING RANDOM NUMBERS; VOTING OR LOTTERY APPARATUS; ARRANGEMENTS, SYSTEMS OR APPARATUS FOR CHECKING NOT PROVIDED FOR ELSEWHERE
    • G07C9/00Individual registration on entry or exit

Definitions

  • This invention consists of the application of the newest technologies in the field of pressure sensors and digital image processing to the measurement of the flow of people coming in and out of a certain area.
  • the operation principle is based on acquiring and analysing pressure images generated by people's footsteps when leaving or going into a public transport vehicle (bus, subway, train, etc.), frequently through a small dimension stair step. And it can also be applied to counting people entering or leaving a certain building, office, shop, etc...
  • the acquired matrixes are analysed using digital image processing algorithms, identifying in this way people's footprints, as well as their direction, allowing a reliable control, regardless the amount of people, footstep speed or any other factor.
  • the obtained data, together with information about the location, date and time of the measurements are saved into a historic database in the same processor.
  • the system here proposed allows an accuracy rated between 96% and 99%, being a cost-effective and easy to use and to install solution.
  • the purpose of this invention is to automatically count the number of persons that pass through an access zone of a certain place or building and to know their direction of movement.
  • the width of the access zone can have any dimension and people can walk through it completely free, one by one or in groups, moving continuously or stopping on it, in one or both directions and with no restrictions on their absolute position on the passing area.
  • the People Counting System by Pressure Images is based on the acquisition of digital pressure images from an existing surface on which people to be counted walk, stepping on it.
  • a pressure image is a bi-dimensional pressure value matrix obtained by scanning the elements that form the sensor surface in a certain moment.
  • the techniques used to measure the pressure exerted on each sensor and the means to scan them are diverse and do not form part of the subject of this patent.
  • the system discriminates between the footprints of several persons stepping simultaneously on the sensor surface. Moreover, if several pressure images per second are provided, the system reliably identifies the direction of movement for each footstep when it leaves the surface and, consequently, the advance direction of the person it belongs to. This is achieved whether the footprints keep completely inside the sensitive surface or whether they imprint only partially on it, as it occurs in the event of a small depth sized step of a train, bus or access stair for a closed place.
  • the People Counting System consists of one or more sensor surfaces which are formed by a bi-dimensional matrix of independent pressure sensible cells, capable of providing a pressure image every few milliseconds; and of one or more pressure image analysing systems capable of collecting the images provided by the sensor surfaces, and, using image processing techniques, detect the partial or complete foot-marks existing on every surface in each scanning period, connect the same foot-marks present on successive scanning periods to identify the footsteps of one person, determine (by their past history) the direction of movement of all of them, link the footprints of each person because the possibility exists that it remains for some time over the sensitive surface, and count people in the right direction when leaving the surface.
  • foot-mark to describe each one of the stepped areas imprinted on the sensitive surface.
  • a footstep of one person may generate, in one scanning instant, several "foot-marks" (areas of active adjacent pixels) that are shown on the pressure image, depending on the number of contacting points on the surface (toe, heel, shoe-side%), on the kind of footwear used and on the cells sensitivity.
  • the number of foot-marks produced by a footstep in the pressure image corresponding to a sampling instant may be higher than two.
  • footstep describes each of the supporting contacts of one person's foot on the sensitive surface, and it has a temporal dimension: a person's footstep is the same one although it may be detected in many successive images, even if the number of foot-marks it leaves on the sensitive surface or its shape or its apparent position in the pressure image changes. Moreover, some of the relevant attributes of the footsteps are their course and advancing direction, magnitudes related to their temporal evolution.
  • the proposed people counting system may be installed in a permanent location or may be placed on-board on a public transport vehicle, since it does not require big size sensitive surfaces and it can be easily adjusted to their access dimensions.
  • This system can also accept other auxiliary information sources, to facilitate both the counting mechanism and the following information treatment.
  • auxiliary information sources are, for instance, a real time clock (to assign temporary attributes to the counts), opening and closing signals from the doors (to avoid erroneous counting when doors are closed or to detect when the equipped vehicle is on a stop), odometer or tachometer signals (to assign spatial or way attributes to the counts, and to identify when the vehicle is moving), data from a GPS receiver (to precisely detect the position of moving systems), signals for pre-determining people's access direction, zero counter signal, etc.
  • a real time clock to assign temporary attributes to the counts
  • opening and closing signals from the doors to avoid erroneous counting when doors are closed or to detect when the equipped vehicle is on a stop
  • odometer or tachometer signals to assign spatial or way attributes to the counts, and to identify when the vehicle is moving
  • data from a GPS receiver to precisely detect the position of moving systems
  • signals for pre-determining people's access direction zero counter signal, etc
  • the obtained data of counted people, and all the other relevant information for each application, are immediately provided and can alternatively be stored in the system, to be used later by an external exploitation system.
  • data storing and communication devices can be added to the People Counting System, as well as suitable information access mechanisms, to facilitate local or remote transmission of data.
  • the People Counting System using Pressure Images is formed by a sensor surface [1] set on an access zone of the control region, plus a pressure image processor [2]; both interconnected by any medium capable of transmitting electric signals [3]; see figure 1.
  • every pressure image captured is converted into a two levels image, each pixel having the value STEPPED or FREE.
  • a graphic representation of a sequence of such images can be seen in figure 3. In it, you can notice the footprints of three different persons printed simultaneously on the sensor surface and its evolution along time. Instants in which people are to be counted as well as their moving direction have been marked using a circle and an arrow.
  • Figure 4 represents the entering image in (t+2) instant, together with the foot-mark list calculated for that image and a table including the appropriate parameters in the same moment (see the explanation of those parameters in the list included at the end of the description).
  • FIG 5 the footstep list originated after processing the images between instant (t) and (t+4) is shown, as well as a graphic representation of the footsteps over the last entering image in moment (t+4).
  • Figure 6 shows the list of people generated after processing images in the same period, this being from (t) to (t+4).
  • the sensor surface is formed by a collection of independent pressure sensors, set as an N row x M column bi-dimensional matrix.
  • N row x M column bi-dimensional matrix the sensor group can be adapted to other polygonal forms by only removing the cells out of the desired perimeter.
  • the image provided by the sensors will be an N x M pixel matrix having zero pressure values in the places corresponding to the cells out of the perimeter.
  • the sensor surface will have its dimensions adapted to the controlled transit zone, being width enough to cover all the stepping zone and with a longitudinal dimension large enough to avoid people skipping it when walking normally, but not as much as to permit several treads of a single person (typically around 40 to 60 cm) although it may step on it with both feet simultaneously. If installed on a stair step, the sensor surface should adapt its dimensions. It is not required for the footprints to be completely inscribed inside the sensitive surface. The size of the surface must be enough to show complete at least the toe or the heel of a footprint, in order to detect in successive images its direction of movement, based on the size and position of the impressed foot-marks. If the sampling rate (in images per second) and the resolution (in pixels per meter) are high enough, the stair step where the system is applied may be as few as 10 or 15 cm. deep longitudinally (in the main direction of people transit).
  • Cell sizes and the distance between them should be as to permit the reproduction of a 5x10cm footprint (as the children's) in not less than 2 x 3 pixels, in order to be able to determine its centre and moving direction. This gives us a maximum sensor distance of about 2.5 cm, being better as it decreases.
  • the sensor surface sensitiveness allows the distinction as different states, of pressure variations as small as a child's weight divided on its two stepping feet; this equals typically a minimum pressure of about 0.2 kg/cm 2 (20 kg standing on two 5 x 10 cm feet). Moreover, the surface is able to receive with no permanent deformation the maximum pressures produced when the whole weight of a person is applied on a slim heel.
  • the sensor response speediness allows the sampling of all the surface to take at least 3 images of footsteps lasting a minimum 100 ms on the surface, what means sending the information at least at a 30 image per second rate (33 ms/image). Furthermore, it responds very quickly to pressure variations, passing from a maximum pressure state to resting position and vice-versa typically in less than 20 ins.
  • the sensor surface is waterproof and resistant to external aggressions as water, soft acids, detergents or soft bases as well as to temperatures in the range of -10° C to +50 ° C. It is also resistant to the erosion and to a certain degree of torsion, flexion, etc. Its external appearance is such as not to attract the attention towards it in any place it is installed (bus, doorway%), in order to avoid people to reject stepping on it and any casual vandalism acts; the existing electronics and the interconnection with the rest of the system remains hidden from people view.
  • pressure values can be transmitted serialised (one data value after another) or in parallel (several data values at a time), in the form of analog (for instance a voltage value which is proportional to each sensor pressure), digital (by means of a number, representing the pressure value of each cell) or binary signals (telling if sensor is activated or not), using a physical medium (metallic or optic fibre) or aerial (radio signals, infrared signals%) or by means of any other state of the art existing electrical signal transmission method. It is noted that the resolution used by the invention, as well as the number of images needed per second, force to use a transmission medium with high capacity and good noise and error protection values.
  • the processor that analyses the acquired pressure images may be a general purpose processor or a specific one for this system. It must fulfil some requirements to be placed on-board on a public transport vehicle (being roughed , low-power consuming, with no moving parts, vibration resistant%), to allow receiving pressure images from several sensitive surfaces, to provide optional permanent data storage and communication capabilities and to accept the connection of those auxiliary devices previously mentioned (real time clock, opening doors signals, odometer or tachometer, GPS location device, one way signal and zero setting signals).
  • the information of each sensible element is transformed into binary format, to allow its further treatment.
  • FIG. 3 shows a graphic presentation of a sequence of pressure images. We can appreciate footprints belonging to three different persons at the same time. The moments in which people are to be counted, as well as their moving direction are marked by a circle and an arrow.
  • the system For each existing image, the system has to identify the boundary zones of the stepped elements. Each of those boundary zones corresponds to the "foot-mark" left by a part of a certain footstep. It must not be a full footprint and there may be a variable number of them (1, 2 or more).
  • Detection of foot-marks is done using just the present binary image, not taking into account the state of the pixels in previous images. Thus, for every sampling period, all the foot-marks on the surface are collected, leaving the calculation of the temporal and motion attributes to the subsequent processing stages. This way the identification of a footstep may be done properly even if it remains a long time over the surface.
  • Figure 4 presents the entering image in moment (t+2), together with the footprint list calculated for the image in that moment.
  • Each footstep generated by the stepping of some person's foot on the sensor surface, may generate several foot-marks on it. Depending on the number of stepping points (toe, heel, shoe-side%) and on the cells sensitivity, one step may result on a group of active zones in the image, close each to other although not adjacent. All the different foot-marks belonging to just one footstep will be set in an image area not larger than a foot maximum dimensions, and set in a longitudinal way along the foot principal's edge.
  • the system uses the footstep's historic information, which keeps current position data (in a similar way to the foot-marks), previous position data (for instance the last boundary rectangle, its centres' list (CA, CGA), its moving direction, its maximum detected dimensions ((XNN,YNN), (XMM, YMM)) and time data (such as the number of images in which it has been present (NI), its appearance instant (TP), the last image in which it has been detected, etc.).
  • current position data in a similar way to the foot-marks
  • previous position data for instance the last boundary rectangle, its centres' list (CA, CGA), its moving direction, its maximum detected dimensions ((XNN,YNN), (XMM, YMM)
  • time data such as the number of images in which it has been present (NI), its appearance instant (TP), the last image in which it has been detected, etc.
  • a certain foot-mark will belong to a certain existing footstep if its centres fall inside the predicted rectangle footstep and its final size, when linking them together, won't overpass the maximum allowed dimensions of a foot.
  • the current and historic footstep's information is updated based on the spatial characteristics of the analysed foot-mark. For instance, the footstep's boundary rectangle gets actualised in the present image, its centres get recalculated, the number of detection or the number of images in which it has appeared, its last appearance time, its maximum dimensions, etc.
  • these data are added to the footstep's information to predict its next possible position and its present moving direction is calculated. To obtain it, the trend of its most recent gravity centre position samples can be used, for example.
  • this stage will provide the list of detected footsteps with new added data using the foot-marks linked to each one on the current image. This list does not disappear with each image, but is stored, keeping the historic information of the footsteps that still are on the sensor surface.
  • the system validates its exit direction. This will be the last moving direction computed when the footstep still had assigned foot-marks. This way, the exit direction can be properly determined for footsteps that have been toggling between the toe an the heel, or from one shoe-side to the other, or whose parts have been detected in the order toe-heel-toe (as it is frequently when going down from a stair step).
  • the foot-marks may be missing in one image, but the footstep may reappear in the following ones
  • its record is removed from the updated footstep list, definitively.
  • Figure 5 shows the footstep list created after processing the images, from instant (t) and (t+4), and a graphic drawing of the footsteps on the last entering image.
  • the goal of this stage is to determine which footsteps correspond to the same person standing on both feet on the sensor surface and which ones belong to different people.
  • the system uses a variety of experimental rules and a set of values to apply them.
  • a list of persons detected on the surface is provided, containing data updated using the footsteps linked to each one in the current image. This list does not disappear with each image but is stored, keeping the historic information of people remaining on the sensor surface.
  • Figure 6 shows a list of people created after processing the images in the same previous interval of time, this is between (t) and (t+4).
  • the system If any person in the list has no more linked footsteps in several consecutive images, the system considers that it has left the sensor surface. It is then counted and deleted from the persons list. The direction of movement given to this person will be the exit direction of its last assigned footstep, which is maintained as a person attribute even when the footsteps disappear.
  • the system has additional information sources, such as door opening signals, vehicle motion detection signals (odometer, tachometer), zero setting or one direction signals, etc. they can be used in this stage to avoid people miscounting in the sensor surface, for instance when doors are closed, or when the vehicle is moving (on board systems) or to predetermine the people counting direction.
  • vehicle motion detection signals odometer, tachometer
  • zero setting or one direction signals etc. they can be used in this stage to avoid people miscounting in the sensor surface, for instance when doors are closed, or when the vehicle is moving (on board systems) or to predetermine the people counting direction.
  • the absolute position of each footstep on the sensor surface is not taken into account, as the clue facts are the spatial and time relations in between the different detected footsteps, and it is contemplated the fact that one or more persons remain for a significantly long time on the sensitive surface. Moreover, they are applicable even if the footstep list comes from one singe platform or from a bigger image formed by several platforms, controlled by one or more processors.
  • This is a flexible method, which adapts to any position of people on the surface not limiting its access zones or its possible movements. Using this system, a typical accuracy of 96.5% in 1,000 people samples can be obtained. The accuracy increases up to 99% in bigger samples (of around 75,000 people).
  • the processor increases the people count until some predetermined event takes place.
  • Such an event may be the generated by the end of the programmed time for periodic storing, the door locking, the movement of the vehicle where the system is installed on, or by pressing the count reset button, etc.
  • the count obtained in each relevant direction is stored in a file, recording the time when the count has been stored and any other significant information (travelled distance, GPS coordinates obtained via satellite, or the status of any input signals).
  • position information is decisive to analyse the people count measures obtained; in fixed systems, the relevant data may be the time of the measured period.
  • the automated system proposed is able to historically store the computed measures and ease its further consulting (communication devices, protocols, etc.). Starting from the data provided by the system, subsequent information can be produced and very useful statistics for any business can be obtained.
  • the sensor surface technology will not cover the whole width of the passing area.
  • several surfaces will be set, one next to the other, until the desired dimensions are obtained.
  • Each sensor surface will provide its own image, as in the general case, but the system will have the particularity that all the surfaces will be synchronised in time by a common processor.
  • the processor will collect the images of all the sensor surfaces and, before starting the analysis, it will attach all of them in memory to form a single image of the necessary dimensions. The remaining processing stages will be done as in the general case.
  • the dimensions of the controlled surface may surpass in connection number, memory capacity, or processing capacity the performance of a single pressure image processor. If this comes to happen, a pyramidal structure formed by several processors linked one to the others is proposed (communicating serially, in parallel, sharing memory). Those processors in the base of the pyramid would acquire the images and treat them until obtaining the footstep lists on the surface, providing its sizes, positions and their remaining characteristics, and sending them to the upper processors. These would collect the information and will do the tracking of footsteps and persons to reach the final counting process and the direction of movement assignment.

Landscapes

  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Measurement Of The Respiration, Hearing Ability, Form, And Blood Characteristics Of Living Organisms (AREA)
  • Indicating Or Recording The Presence, Absence, Or Direction Of Movement (AREA)

Abstract

This people counting system is composed of one or more sensor surfaces each one formed by independent pressure sensible cells in a bi-dimensional matrix arrangement, and frequently installed on a small dimension stair step, capable of providing a pressure image every few milliseconds; and of one or more pressure image analysing systems capable of collecting the images generated through the sensor surfaces and, by means of image processing methods, detect the number of footprints left on every surface during each sampling period, link the footprints detected on consecutive events, determine (by their past history) their moving direction, group the possible footprints of each person and count them on the correct direction when they leave the surface.

Description

  • This invention consists of the application of the newest technologies in the field of pressure sensors and digital image processing to the measurement of the flow of people coming in and out of a certain area. The operation principle is based on acquiring and analysing pressure images generated by people's footsteps when leaving or going into a public transport vehicle (bus, subway, train, etc.), frequently through a small dimension stair step. And it can also be applied to counting people entering or leaving a certain building, office, shop, etc...
  • Combining the two above mentioned technologies, a strong, reliable, low cost, versatile and installable on-board in a mass transportation vehicle, people counting system is proposed.
  • By means of an adequate sensor, surface bi-dimensional images containing information about the pressure applied on the surface when people walk on it are obtained. These images are real-time generated and with a relatively high resolution, for pressure elements.
  • Then, the acquired matrixes are analysed using digital image processing algorithms, identifying in this way people's footprints, as well as their direction, allowing a reliable control, regardless the amount of people, footstep speed or any other factor. The obtained data, together with information about the location, date and time of the measurements are saved into a historic database in the same processor.
  • In order to analyse the generated data, they can be recovered using an operational sub-system designed for this purpose, which also allows to work on the statistic evaluation tasks, graphic presentation and to export the data to other remote analysing systems.
  • State of the art
  • The technical development of people or passenger counting systems has been stopped world-wide for several years, because of the impossibility of obtaining cost-effective high technology solutions, until very recent times. In this sense, we can count the US Patents US-4.00.400, US-4.122.331, US-4.175.446, US-4.303.851, WO-97/22089 and WO-97/23376 in which counting systems are basically based on detecting, in a more or less complicated way, the footsteps on the bus stairs and thereafter, using simple algorithms, count the number of passengers in the bus. Although, in some case, inventions are more complicated, as the last one mentioned above, which is similar in its external aspects and applications to the one here described.
  • Being the number of used footstep sensors very low (one by step or two in several steps in the most complex systems), the error rate grows very rapidly, specially in crowded situations, which are the usual ones. Moreover, the existence of steps is often necessary for the use of these inventions.
  • Other recent solutions are closer to this proposal, even using bi-dimensional arrays of pressure sensitive sensors and processing of footprints to compute the number of persons walking on the access surface and to determine their moving direction. One of these is, for instance, the publication in WO 98/40719, although it claims for a method instead of an entire system, and it does not solve problems such as static or incomplete footprints. That makes this method hardly applicable for environments as entrances of mass transportation systems.
  • On the other hand, there is a number of patents that claim for constructing methods to manufacture pressure sensitive surfaces but that subject is omitted in this patent.
  • Other technologies use infrared sensors installed on the ceiling of the vehicle, using multiple sensors for each door. Precision on the most simple systems can't overpass the 75% for one-direction traffic. The most complex and expensive systems can guarantee an accuracy of up to 90 % even in bi-directional traffic conditions.
  • The system here proposed allows an accuracy rated between 96% and 99%, being a cost-effective and easy to use and to install solution.
  • General description of the invention
  • The purpose of this invention is to automatically count the number of persons that pass through an access zone of a certain place or building and to know their direction of movement. The width of the access zone can have any dimension and people can walk through it completely free, one by one or in groups, moving continuously or stopping on it, in one or both directions and with no restrictions on their absolute position on the passing area.
  • Supposing any way of going in or out of a specific building, public transport vehicle or any other closed place which has walls and access zones (for our purpose it does not matter that people come in and out in an ordered way or not), it is a must that people step on the floor (or on any existing steps) to move in or out of this certain place. To do so they will have to step on a certain limited surface of the access zone, or on a step if stairs exist. We will initially suppose that both the probability of jumping over these zones and the probability of advancing with successive steps across them are very low (due to their design and dimensions) and that these events are not very relevant.
  • The People Counting System by Pressure Images is based on the acquisition of digital pressure images from an existing surface on which people to be counted walk, stepping on it. A pressure image is a bi-dimensional pressure value matrix obtained by scanning the elements that form the sensor surface in a certain moment. The techniques used to measure the pressure exerted on each sensor and the means to scan them are diverse and do not form part of the subject of this patent.
  • Using matrix sensors with an appropriate spatial resolution, the system discriminates between the footprints of several persons stepping simultaneously on the sensor surface. Moreover, if several pressure images per second are provided, the system reliably identifies the direction of movement for each footstep when it leaves the surface and, consequently, the advance direction of the person it belongs to. This is achieved whether the footprints keep completely inside the sensitive surface or whether they imprint only partially on it, as it occurs in the event of a small depth sized step of a train, bus or access stair for a closed place.
  • The People Counting System consists of one or more sensor surfaces which are formed by a bi-dimensional matrix of independent pressure sensible cells, capable of providing a pressure image every few milliseconds; and of one or more pressure image analysing systems capable of collecting the images provided by the sensor surfaces, and, using image processing techniques, detect the partial or complete foot-marks existing on every surface in each scanning period, connect the same foot-marks present on successive scanning periods to identify the footsteps of one person, determine (by their past history) the direction of movement of all of them, link the footprints of each person because the possibility exists that it remains for some time over the sensitive surface, and count people in the right direction when leaving the surface.
  • In the context of this description, we will be using the term "foot-mark" to describe each one of the stepped areas imprinted on the sensitive surface. A footstep of one person may generate, in one scanning instant, several "foot-marks" (areas of active adjacent pixels) that are shown on the pressure image, depending on the number of contacting points on the surface (toe, heel, shoe-side...), on the kind of footwear used and on the cells sensitivity. Specifically the number of foot-marks produced by a footstep in the pressure image corresponding to a sampling instant may be higher than two.
  • The term "footstep" describes each of the supporting contacts of one person's foot on the sensitive surface, and it has a temporal dimension: a person's footstep is the same one although it may be detected in many successive images, even if the number of foot-marks it leaves on the sensitive surface or its shape or its apparent position in the pressure image changes. Moreover, some of the relevant attributes of the footsteps are their course and advancing direction, magnitudes related to their temporal evolution.
  • The proposed people counting system may be installed in a permanent location or may be placed on-board on a public transport vehicle, since it does not require big size sensitive surfaces and it can be easily adjusted to their access dimensions.
  • This system can also accept other auxiliary information sources, to facilitate both the counting mechanism and the following information treatment. Such sources are, for instance, a real time clock (to assign temporary attributes to the counts), opening and closing signals from the doors (to avoid erroneous counting when doors are closed or to detect when the equipped vehicle is on a stop), odometer or tachometer signals (to assign spatial or way attributes to the counts, and to identify when the vehicle is moving), data from a GPS receiver (to precisely detect the position of moving systems), signals for pre-determining people's access direction, zero counter signal, etc.
  • The obtained data of counted people, and all the other relevant information for each application, are immediately provided and can alternatively be stored in the system, to be used later by an external exploitation system. To meet this requirement, data storing and communication devices can be added to the People Counting System, as well as suitable information access mechanisms, to facilitate local or remote transmission of data.
  • The main advantages of this system compared to those existing in the same field are:
    • its accuracy, thanks to its capability of detecting different footprints belonging to different people stepping at the same time on the transit sensor surface and therefore minimizing the "occlusions" of other technologies (infrared barriers, video cameras, only longitudinal or transversal sensors);
    • it is an intrinsically bi-directional counting method, as the analysis of a high number of images per second allows to recognise the form and position of each footstep along time;
    • it goes unnoticed to the general public, reducing the risk of vandalism and the possibility that detection affects the obtained measures;
    • it does not block people passing, easing the access and evacuation of the places whose capacity wants to be measured;
    • the small size of the sensitive surface needed, that decreases their own cost, reduces the processing capacity requirements for the corresponding pressure images and minimises the counting errors due to the identification of successive strides of the same person, most of all in crowded areas;
    • the proper handling of footsteps belonging to people who stay a prolonged time over the surface, because of queues or obstacles obstructing the continuous movement of persons.
    Brief explanation of the drawings.
  • In its most basic version, the People Counting System using Pressure Images is formed by a sensor surface [1] set on an access zone of the control region, plus a pressure image processor [2]; both interconnected by any medium capable of transmitting electric signals [3]; see figure 1.
  • In the processor, images transmitted by the sensor surface are captured and, for each one, the following processing stages are carried out, as seen in figure 2: once the pressure images IP have been acquired, the processor will do a binary conversion (4) of them; using the binary images IB, a classification of the stepped zones (5) is made, and, based on this, an LHI image foot-mark list is built. Thereinafter, an assignment of foot-marks to footsteps (6) is made, using for this purpose the present foot-mark list (LHI) and the previous footstep list, LI; therefore an updated footstep list LIA is obtained. After this procedure and based on a collection of rules (LR) specifically designed to this goal, footsteps are grouped in persons (7), using the present footstep list and the previous person list, obtained in the preceding image; thus, the updated people list, LEA, for the present image is calculated. Finally, adding several convenient additional signals (SA), the system counts people (8), resulting in a certain number of persons moving in each direction, NPCS.
  • As a result of the pressure images binary conversion stage, every pressure image captured is converted into a two levels image, each pixel having the value STEPPED or FREE. A graphic representation of a sequence of such images can be seen in figure 3. In it, you can notice the footprints of three different persons printed simultaneously on the sensor surface and its evolution along time. Instants in which people are to be counted as well as their moving direction have been marked using a circle and an arrow.
  • Figure 4 represents the entering image in (t+2) instant, together with the foot-mark list calculated for that image and a table including the appropriate parameters in the same moment (see the explanation of those parameters in the list included at the end of the description).
  • In figure 5 the footstep list originated after processing the images between instant (t) and (t+4) is shown, as well as a graphic representation of the footsteps over the last entering image in moment (t+4).
  • Figure 6 shows the list of people generated after processing images in the same period, this being from (t) to (t+4).
  • Detailed description and preferred embodiments
  • The sensor surface is formed by a collection of independent pressure sensors, set as an N row x M column bi-dimensional matrix. To begin, we will suppose that we have a rectangular surface, although the sensor group can be adapted to other polygonal forms by only removing the cells out of the desired perimeter. In any case, the image provided by the sensors will be an N x M pixel matrix having zero pressure values in the places corresponding to the cells out of the perimeter.
  • The sensor surface will have its dimensions adapted to the controlled transit zone, being width enough to cover all the stepping zone and with a longitudinal dimension large enough to avoid people skipping it when walking normally, but not as much as to permit several treads of a single person (typically around 40 to 60 cm) although it may step on it with both feet simultaneously. If installed on a stair step, the sensor surface should adapt its dimensions. It is not required for the footprints to be completely inscribed inside the sensitive surface. The size of the surface must be enough to show complete at least the toe or the heel of a footprint, in order to detect in successive images its direction of movement, based on the size and position of the impressed foot-marks. If the sampling rate (in images per second) and the resolution (in pixels per meter) are high enough, the stair step where the system is applied may be as few as 10 or 15 cm. deep longitudinally (in the main direction of people transit).
  • Cell sizes and the distance between them should be as to permit the reproduction of a 5x10cm footprint (as the children's) in not less than 2 x 3 pixels, in order to be able to determine its centre and moving direction. This gives us a maximum sensor distance of about 2.5 cm, being better as it decreases.
  • The sensor surface sensitiveness allows the distinction as different states, of pressure variations as small as a child's weight divided on its two stepping feet; this equals typically a minimum pressure of about 0.2 kg/cm2 (20 kg standing on two 5 x 10 cm feet). Moreover, the surface is able to receive with no permanent deformation the maximum pressures produced when the whole weight of a person is applied on a slim heel.
  • The sensor response speediness allows the sampling of all the surface to take at least 3 images of footsteps lasting a minimum 100 ms on the surface, what means sending the information at least at a 30 image per second rate (33 ms/image). Furthermore, it responds very quickly to pressure variations, passing from a maximum pressure state to resting position and vice-versa typically in less than 20 ins.
  • The sensor surface is waterproof and resistant to external aggressions as water, soft acids, detergents or soft bases as well as to temperatures in the range of -10° C to +50 ° C. It is also resistant to the erosion and to a certain degree of torsion, flexion, etc. Its external appearance is such as not to attract the attention towards it in any place it is installed (bus, doorway...), in order to avoid people to reject stepping on it and any casual vandalism acts; the existing electronics and the interconnection with the rest of the system remains hidden from people view.
  • Every pressure image obtained on the sensor surface is real time sent to the processor over different interconnecting devices, what is achieved using an open interconnection interface. Thus, pressure values can be transmitted serialised (one data value after another) or in parallel (several data values at a time), in the form of analog (for instance a voltage value which is proportional to each sensor pressure), digital (by means of a number, representing the pressure value of each cell) or binary signals (telling if sensor is activated or not), using a physical medium (metallic or optic fibre) or aerial (radio signals, infrared signals...) or by means of any other state of the art existing electrical signal transmission method. It is noted that the resolution used by the invention, as well as the number of images needed per second, force to use a transmission medium with high capacity and good noise and error protection values.
  • The processor that analyses the acquired pressure images may be a general purpose processor or a specific one for this system. Anyway it must fulfil some requirements to be placed on-board on a public transport vehicle (being roughed , low-power consuming, with no moving parts, vibration resistant...), to allow receiving pressure images from several sensitive surfaces, to provide optional permanent data storage and communication capabilities and to accept the connection of those auxiliary devices previously mentioned (real time clock, opening doors signals, odometer or tachometer, GPS location device, one way signal and zero setting signals...).
  • As the incoming images transmitted by the sensor surface are being captured, the following stages are carried out in the processor for every sampling period:
  • - Pressure image binary conversion (4):
  • Regardless of the pressure image transmission means, the information of each sensible element is transformed into binary format, to allow its further treatment. In the case of analog or digital images, this would be done by comparing the pressure level or value with a reference level which can be a single one for the whole surface or a particular one for each cell depending on its previous state. The elements surpassing the reference level will be active (stepped) and those which don't overpass it will be in resting state.
  • As a result of this first stage, every pressure image captured is transformed into a two-level image, having now each pixel an STEPPED/NOT-STEPPED value. Figure 3 shows a graphic presentation of a sequence of pressure images. We can appreciate footprints belonging to three different persons at the same time. The moments in which people are to be counted, as well as their moving direction are marked by a circle and an arrow.
  • - Stepped zone grouping (5):
  • For each existing image, the system has to identify the boundary zones of the stepped elements. Each of those boundary zones corresponds to the "foot-mark" left by a part of a certain footstep. It must not be a full footprint and there may be a variable number of them (1, 2 or more).
  • Detection of foot-marks is done using just the present binary image, not taking into account the state of the pixels in previous images. Thus, for every sampling period, all the foot-marks on the surface are collected, leaving the calculation of the temporal and motion attributes to the subsequent processing stages. This way the identification of a footstep may be done properly even if it remains a long time over the surface.
  • Amongst the different image processing techniques, several methods can be used in order to identity and characterise those foot-marks depending on the connectivity of the active pixels, as well as different noise filtering algorithms (to remove falsely detected pixels) which could have been generated in the previous stage.
  • These techniques allow the system to identify, for each boundary zone or foot-mark, a set of spatial attributes, such as the coordinates of the rectangle that inscribes it (XN, YN), (XM,YM), its different centres (geometrical centre (C), gravity centre (G), or other), the number of active points which form it (N), a boundary polygon, etc.
  • As a result of this stage, a list of foot-marks or stepped zones is created, each one together with its own characteristics.
  • Figure 4 presents the entering image in moment (t+2), together with the footprint list calculated for the image in that moment.
  • - Foot-mark to footstep linkage (6):
  • Each footstep, generated by the stepping of some person's foot on the sensor surface, may generate several foot-marks on it. Depending on the number of stepping points (toe, heel, shoe-side...) and on the cells sensitivity, one step may result on a group of active zones in the image, close each to other although not adjacent. All the different foot-marks belonging to just one footstep will be set in an image area not larger than a foot maximum dimensions, and set in a longitudinal way along the foot principal's edge.
  • To decide on every image, which foot-marks correspond to each footstep, the system uses the footstep's historic information, which keeps current position data (in a similar way to the foot-marks), previous position data (for instance the last boundary rectangle, its centres' list (CA, CGA), its moving direction, its maximum detected dimensions ((XNN,YNN), (XMM, YMM)) and time data (such as the number of images in which it has been present (NI), its appearance instant (TP), the last image in which it has been detected, etc.).
  • This way there is no need of storing and re-processing neither previous images nor the foot-mark lists from all the preceding instants. Processing is always done based on information elaborated in previous images plus that obtained from the present one, not depending on retroactive computation.
  • A certain foot-mark will belong to a certain existing footstep if its centres fall inside the predicted rectangle footstep and its final size, when linking them together, won't overpass the maximum allowed dimensions of a foot. When assigning a foot-mark to a certain footstep, the current and historic footstep's information is updated based on the spatial characteristics of the analysed foot-mark. For instance, the footstep's boundary rectangle gets actualised in the present image, its centres get recalculated, the number of detection or the number of images in which it has appeared, its last appearance time, its maximum dimensions, etc. Once the analysis of all the image's foot-marks has finished, these data are added to the footstep's information to predict its next possible position and its present moving direction is calculated. To obtain it, the trend of its most recent gravity centre position samples can be used, for example.
  • If no footstep to which associate the current print is found, a new footstep will be created and the foot-mark will be linked to it, initialising its data according to the current footprint's characteristics.
  • As an exit, this stage will provide the list of detected footsteps with new added data using the foot-marks linked to each one on the current image. This list does not disappear with each image, but is stored, keeping the historic information of the footsteps that still are on the sensor surface.
  • When any footstep in the list has no longer a foot-mark linked to it, the system validates its exit direction. This will be the last moving direction computed when the footstep still had assigned foot-marks. This way, the exit direction can be properly determined for footsteps that have been toggling between the toe an the heel, or from one shoe-side to the other, or whose parts have been detected in the order toe-heel-toe (as it is frequently when going down from a stair step). It has been experimentally probed that, whatever the sequence of contact parts of a foot has been, while it is staying over the sensitive surface, practically the whole people raise their feet from heel to toe when they leave the surface; and that it is much more reliable to use the apparent advancing direction of one footprint when it is leaving the pressure images than when it newly appears on the surface to determine its moving direction.
  • If the lack of associated foot-marks condition is maintained for a footstep during several images (the foot-marks may be missing in one image, but the footstep may reappear in the following ones), its record is removed from the updated footstep list, definitively.
  • Figure 5 shows the footstep list created after processing the images, from instant (t) and (t+4), and a graphic drawing of the footsteps on the last entering image.
  • - Footsteps to persons assignment (7):
  • The goal of this stage is to determine which footsteps correspond to the same person standing on both feet on the sensor surface and which ones belong to different people.
  • The system uses a variety of experimental rules and a set of values to apply them.
  • These data are adjusted after experimental analysis in "standard applications" (a bus or train entrance stair step, the entrance of a mall or of a certain store, etc.) where those values are obtained using statistical methods to produce the best precision results in the final count of people.
  • Some of the rules to group footsteps into persons are the following:
    • Footsteps which centres are not longer from a certain distance may be grouped, being this distance the normal distance between a person's feet, and which must be adjusted statistically in previous tests.
    • It is not allowed to add more than two footsteps to one person.
    • A footstep detected very few times is not allowed to group with any other one (because there is no reliable information on it). The minimum number of detections to link one footstep to other is experimentally determined for each "standard installation".
    • Two footsteps with different moving direction (in between the possible established) or which moving vectors form an angle greater than a predetermined value can not be grouped together.
    • Two footsteps having another one in between them, not compatible with any of them, cannot be grouped together.
    • If the time of the first detection of two footsteps appearing on the sensor surface is lower than a predetermined experimental value, the steps can not be grouped together (nobody enters the surface with both feet at the same time).
    • A footstep that has just disappeared (in a previous moment) can only be grouped with another one in the same position, and will belong to the same person's foot. If a new footstep appears moved from its previous position it must belong to a different person.
  • Applying each of these rules and some more which can be fixed is subject to the tests carried out in the "standard installations".
  • As an exit of this stage a list of persons detected on the surface is provided, containing data updated using the footsteps linked to each one in the current image. This list does not disappear with each image but is stored, keeping the historic information of people remaining on the sensor surface.
  • People processing is always done based on the person list information elaborated in previous images plus the footstep list calculated from the current one. This way there is no need of storing and re-processing neither previous images nor the footstep lists from all the preceding instants.
  • Figure 6 shows a list of people created after processing the images in the same previous interval of time, this is between (t) and (t+4).
  • - People counting (8):
  • If any person in the list has no more linked footsteps in several consecutive images, the system considers that it has left the sensor surface. It is then counted and deleted from the persons list. The direction of movement given to this person will be the exit direction of its last assigned footstep, which is maintained as a person attribute even when the footsteps disappear.
  • If the system has additional information sources, such as door opening signals, vehicle motion detection signals (odometer, tachometer), zero setting or one direction signals, etc. they can be used in this stage to avoid people miscounting in the sensor surface, for instance when doors are closed, or when the vehicle is moving (on board systems) or to predetermine the people counting direction.
  • In all these processing stages, the absolute position of each footstep on the sensor surface is not taken into account, as the clue facts are the spatial and time relations in between the different detected footsteps, and it is contemplated the fact that one or more persons remain for a significantly long time on the sensitive surface. Moreover, they are applicable even if the footstep list comes from one singe platform or from a bigger image formed by several platforms, controlled by one or more processors. This is a flexible method, which adapts to any position of people on the surface not limiting its access zones or its possible movements. Using this system, a typical accuracy of 96.5% in 1,000 people samples can be obtained. The accuracy increases up to 99% in bigger samples (of around 75,000 people).
  • The processor increases the people count until some predetermined event takes place. Such an event may be the generated by the end of the programmed time for periodic storing, the door locking, the movement of the vehicle where the system is installed on, or by pressing the count reset button, etc. Then the count obtained in each relevant direction is stored in a file, recording the time when the count has been stored and any other significant information (travelled distance, GPS coordinates obtained via satellite, or the status of any input signals).
  • For on-board systems, position information is decisive to analyse the people count measures obtained; in fixed systems, the relevant data may be the time of the measured period. In both kind of systems the automated system proposed is able to historically store the computed measures and ease its further consulting (communication devices, protocols, etc.). Starting from the data provided by the system, subsequent information can be produced and very useful statistics for any business can be obtained.
  • Other system implementations allow the connection of several sensor surfaces, each corresponding to one of the existing entrances to a certain place, to just one processor, having different inputs to this effect. In this case, the image analysing and people count procedure is repeated for each image of each surface independently. The global bi-directional people scores may be collected in order to obtain the total number of visits to the controlled place.
  • In other applications, it is possible that the sensor surface technology will not cover the whole width of the passing area. In this case, several surfaces will be set, one next to the other, until the desired dimensions are obtained. Each sensor surface will provide its own image, as in the general case, but the system will have the particularity that all the surfaces will be synchronised in time by a common processor. The processor will collect the images of all the sensor surfaces and, before starting the analysis, it will attach all of them in memory to form a single image of the necessary dimensions. The remaining processing stages will be done as in the general case.
  • On the other hand, the dimensions of the controlled surface may surpass in connection number, memory capacity, or processing capacity the performance of a single pressure image processor. If this comes to happen, a pyramidal structure formed by several processors linked one to the others is proposed (communicating serially, in parallel, sharing memory...). Those processors in the base of the pyramid would acquire the images and treat them until obtaining the footstep lists on the surface, providing its sizes, positions and their remaining characteristics, and sending them to the upper processors. These would collect the information and will do the tracking of footsteps and persons to reach the final counting process and the direction of movement assignment.
  • LIST OF PARAMETERS USED IN FIGURES 4, 5 AND 6
  • Figure 4:
    • H1, H2, H3, H4, H5, H6, and H7 represent the different foot-marks left by three persons footsteps on the sensor surface, in moment (t+2).
    • XM, XN, YM and YN are the maximum and minimum X and Y coordinates of each foot-mark, according to a theoretical coordinate system on the surface.
    • C represents geometrical centre of a foot-mark.
    • CG is the foot-mark gravity centre (ΣXi/N, ΣYi/N)
    • N is the number of active cells.
  • Figure 5:
    • P1, P2, P3 and P4 are the generated footsteps after processing images from instant (t) to (t+4).
    • CA represents the previous geometrical centres of the footstep.
    • CGA represent the previous gravity centres of the footstep.
    • XNN, XMM, YNN, YMM are the minimum and maximum values of the XN, XM, YN and YM.
    • NI is the number of images in which a certain footstep has been detected.
    • TP is the moment in which the footstep firstly appeared.
  • Figure 6:
    • PS1, PS2 and PS3 represent people detected after processing the footstep in the images.
    • CONT indicates if the person has been counted or not.
    • SAP indicated the current moving direction associated to the person (it will be definitive when the footstep exits the surface): AR= up and AB= down.

Claims (11)

  1. People counting system using pressure images comprising one or more sensor surfaces and one or more processors, linked to each other by any signal transmission medium, wherein each one of the sensor surfaces is formed by a bi-dimensional array of independent cells sensible to the pressure; from such surfaces, in consecutive sampling times, digital pressure images are obtained of the whole or partial footprints left by people walking on it, these images being analysed by image processing techniques to count the people passing over the surface, not being a requirement for the mentioned surface to completely contain an entire footprint of a person.
  2. People counting system using pressure images, according to claim 1, wherein one or more sensor surfaces cover the accesses to a particular place and their images are analysed by one or more processors, counting in real time the people passing over the surfaces and assigning their moving direction when they leave them, with no practical restriction regarding the number of people on the surfaces or regarding their absolute position on them, not even on the possibility that people remain a finite time on the surfaces.
  3. People counting system using pressure images, according to any of the previous claims, wherein the interface between the sensor surfaces and the image processor, allows the transmission of the pressure values serially or in parallel, in the form of analog, digital or binary signals, over a physical or aerial medium or any other electric signal transmitting method.
  4. People counting system using pressure images, according to any of the previous claims, wherein the processor, once collected and binary convened the sensor surface images, detects the existing foot-marks in each sampling period, relates the different foot-marks each to others to identify partial or complete footsteps, calculates the current moving direction of the footsteps (based on the detected foot-marks for the present image and on information elaborated in previous sampling instants), links the footsteps of each person and counts it in the appropriate direction when it leaves the surface.
  5. People counting system using pressure images, according to any of the previous claims, wherein the response time of the sensor allows taking at least 3 samples of footsteps remaining on the surface at least 100 ms, which equals sending the information at least 30 images per second.
  6. People counting system using pressure images, according to any of the previous claims, wherein for each individual acquired image, disregarding any other ones, the stepped zones are grouped and a list containing spatial information about the foot-marks is created.
  7. People counting system using pressure images, according to any of the previous claims, wherein the foot-marks are assigned to footsteps, creating a list of the footsteps detected on the surface, and to decide which marks belong to a footstep, with no limit concerning their number, historic previous information about the footsteps and spatial data of the foot-marks in the current image are used.
  8. People counting system using pressure images, according to any of the previous claims, wherein the footsteps of the list which have no assigned foot-marks get their exit direction validated, previously calculated using their most recent positioning data as it leaves contact with the sensitive surface and, when this condition remains true for several images, are then deleted from the footstep list.
  9. People counting system using pressure images, according to any of the previous claims, wherein the footsteps are grouped, forming then a person list staying on the surface for each sampling period, using a collection of particular rules.
  10. People counting system using pressure images, according to any of the previous claims, wherein for any of the persons in the list which have no assigned footsteps during several consecutive images, it is supposed to have left the sensor surface, it is then counted and deleted from the people list. The exit direction attributed to it is the exit direction of the last of its associated footsteps.
  11. People counting system using pressure images, according to any of the previous claims, wherein additional inputs are provided, such as door opening signals, motion indication signals for systems installed on-board vehicles, reset counter signals or one way signals, which are used to determine the passing direction of the counted people and/or to avoid miscounting, although the system operation is not restricted by their condition.
EP99500166A 1998-09-22 1999-09-17 People counting system by pressure images Withdrawn EP0992949A3 (en)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
ES9801987 1998-09-22
ES9801987A ES2151844B1 (en) 1998-09-22 1998-09-22 PEOPLE ACCOUNT SYSTEM FOR PRESSURE IMAGES

Publications (2)

Publication Number Publication Date
EP0992949A2 true EP0992949A2 (en) 2000-04-12
EP0992949A3 EP0992949A3 (en) 2001-05-16

Family

ID=8305236

Family Applications (1)

Application Number Title Priority Date Filing Date
EP99500166A Withdrawn EP0992949A3 (en) 1998-09-22 1999-09-17 People counting system by pressure images

Country Status (2)

Country Link
EP (1) EP0992949A3 (en)
ES (1) ES2151844B1 (en)

Cited By (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002041262A1 (en) * 2000-11-16 2002-05-23 Laurence Haquet System for analysing people's movements
FR2831673A1 (en) * 2001-10-30 2003-05-02 Volume Contact PASSAGE DETECTION DEVICE
DE102004057000B3 (en) * 2004-11-25 2006-06-08 Siemens Ag Emergency and evacuation system for e.g. hotel, has pressure sensors arranged in groups/belts that are distributed in large object such that distribution and running direction of persons are determined from signals sent by sensors to center
EP2207128A1 (en) 2009-01-08 2010-07-14 Robert Bosch GmbH Monitoring device and sensor device for recording the floor contact areas of a monitoring object for the monitoring device
DE102005048647B4 (en) * 2005-10-11 2011-03-17 Wvi Prof. Dr. Wermuth Verkehrsforschung Und Infrastrukturplanung Gmbh Procedure for determining the number of passengers on public transport
CN103198327A (en) * 2013-04-25 2013-07-10 重庆师范大学 Open type scene-oriented entrance and exit population counting method and device
CN103198548A (en) * 2013-04-03 2013-07-10 深圳职业技术学院 Regional bus on-off passenger number distinguishing algorithm based on switch sensors
CN103235932A (en) * 2013-04-09 2013-08-07 重庆理工大学 People counting method and device for accesses
US9672462B2 (en) 2015-03-03 2017-06-06 Umm Al-Qura University Smart surface-mounted hybrid sensor system, method, and apparatus for counting
GB2545003A (en) * 2015-12-03 2017-06-07 Int Cons Airlines Group Queue monitoring based on imprint profiles
CN109323785A (en) * 2018-11-28 2019-02-12 浙江科技学院 Subway carriage crowding shifts to an earlier date predictor and its forecasting procedure
US10249013B2 (en) 2015-02-03 2019-04-02 Alibaba Group Holding Limited Method and system for wireless payment of public transport fare
US10248954B2 (en) 2014-08-14 2019-04-02 Alibaba Group Holding Limited Method and system for verifying user identity using card features
JP2019061318A (en) * 2017-09-25 2019-04-18 東洋インキScホールディングス株式会社 Moving direction specification method and pressure-sensitive sensor device
US10268166B2 (en) 2016-09-15 2019-04-23 Otis Elevator Company Intelligent surface systems for building solutions
US10275813B2 (en) 2014-07-08 2019-04-30 Alibaba Group Holding Limited Method and system for providing a transaction platform for pre-owned merchandise
US10296636B2 (en) 2015-10-09 2019-05-21 Alibaba Group Holding Limited Efficient navigation category management
US10325088B2 (en) 2014-07-03 2019-06-18 Alibaba Group Holding Limited Method and system for information authentication
US10579973B2 (en) 2015-01-19 2020-03-03 Alibaba Group Holding Limited System for efficient processing of transaction requests related to an account in a database
US10755345B2 (en) 2014-12-03 2020-08-25 Alibaba Group Holding Limited System and method for secure account transfer
CN112907785A (en) * 2021-01-15 2021-06-04 广东趣买票科技有限公司 Scenic spot number control system and control device based on Internet of things
US11538039B2 (en) 2018-02-12 2022-12-27 Advanced New Technologies Co., Ltd. Method and system for facilitating risk control of an online financial platform
DE102021122969A1 (en) 2021-09-06 2023-03-09 Incoretex GmbH Method for the continuous monitoring of an occupancy of a passenger transport vehicle
US11816714B2 (en) 2018-03-19 2023-11-14 Advanced New Technologies Co., Ltd. Service verification method and apparatus

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113140092A (en) * 2021-04-13 2021-07-20 云南云能科技有限公司 System and method for monitoring personnel in public place

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US400400A (en) 1889-03-26 gtjillaumin
US4122331A (en) 1977-06-01 1978-10-24 Giken Trading Company Passer counter
US4175446A (en) 1978-04-26 1979-11-27 The University Of Iowa Research Foundation Step counting device and method
US4303851A (en) 1979-10-16 1981-12-01 Otis Elevator Company People and object counting system
WO1997022089A1 (en) 1994-06-08 1997-06-19 Brime S.A. Device for detecting the presence and direction of passage of mobiles and persons with the view to counting them
WO1997023376A1 (en) 1995-12-21 1997-07-03 Mafiss Ltd. Automatic counting of people
WO1998040719A1 (en) 1997-03-10 1998-09-17 Aharon Shapira Method for counting the number of people crossing an entry barrier

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JPH03196286A (en) * 1989-12-25 1991-08-27 Mirai Biru Kenkyu Kaihatsu Kk Detector for number of persons in building
NL9100591A (en) * 1991-04-05 1992-11-02 Has Transport Systems B V Pressure sensitive mat counting feet to monitor passenger traffic - is placed in entry or exit path of vehicle, building etc. to detect and count persons passing through
JPH05324955A (en) * 1992-05-27 1993-12-10 Kazuo Nakazawa Device for counting the number of persons
JPH06309536A (en) * 1993-04-26 1994-11-04 Ace Denken:Kk Passer counting device
FR2713805B1 (en) * 1993-12-15 1996-09-06 Alkan Sa Anti-fraud system for public transport use.
IL114440A0 (en) * 1995-07-04 1995-11-27 Shapira Aharon Method and system for estimating an available surface area

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US400400A (en) 1889-03-26 gtjillaumin
US4122331A (en) 1977-06-01 1978-10-24 Giken Trading Company Passer counter
US4175446A (en) 1978-04-26 1979-11-27 The University Of Iowa Research Foundation Step counting device and method
US4303851A (en) 1979-10-16 1981-12-01 Otis Elevator Company People and object counting system
WO1997022089A1 (en) 1994-06-08 1997-06-19 Brime S.A. Device for detecting the presence and direction of passage of mobiles and persons with the view to counting them
WO1997023376A1 (en) 1995-12-21 1997-07-03 Mafiss Ltd. Automatic counting of people
WO1998040719A1 (en) 1997-03-10 1998-09-17 Aharon Shapira Method for counting the number of people crossing an entry barrier

Cited By (32)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2002041262A1 (en) * 2000-11-16 2002-05-23 Laurence Haquet System for analysing people's movements
FR2831673A1 (en) * 2001-10-30 2003-05-02 Volume Contact PASSAGE DETECTION DEVICE
EP1308905A1 (en) * 2001-10-30 2003-05-07 Volume Contact SARL Passage detection apparatus
DE102004057000B3 (en) * 2004-11-25 2006-06-08 Siemens Ag Emergency and evacuation system for e.g. hotel, has pressure sensors arranged in groups/belts that are distributed in large object such that distribution and running direction of persons are determined from signals sent by sensors to center
DE102005048647B4 (en) * 2005-10-11 2011-03-17 Wvi Prof. Dr. Wermuth Verkehrsforschung Und Infrastrukturplanung Gmbh Procedure for determining the number of passengers on public transport
EP2207128A1 (en) 2009-01-08 2010-07-14 Robert Bosch GmbH Monitoring device and sensor device for recording the floor contact areas of a monitoring object for the monitoring device
CN103198548B (en) * 2013-04-03 2015-07-29 深圳职业技术学院 Bus based on switch sensor upper and lower ridership area judging algorithm
CN103198548A (en) * 2013-04-03 2013-07-10 深圳职业技术学院 Regional bus on-off passenger number distinguishing algorithm based on switch sensors
CN103235932A (en) * 2013-04-09 2013-08-07 重庆理工大学 People counting method and device for accesses
CN103235932B (en) * 2013-04-09 2016-06-08 重庆理工大学 Gateway demographic method and device
CN103198327B (en) * 2013-04-25 2016-12-28 重庆师范大学 Gateway demographic method and device towards open scene
CN103198327A (en) * 2013-04-25 2013-07-10 重庆师范大学 Open type scene-oriented entrance and exit population counting method and device
US10325088B2 (en) 2014-07-03 2019-06-18 Alibaba Group Holding Limited Method and system for information authentication
US10275813B2 (en) 2014-07-08 2019-04-30 Alibaba Group Holding Limited Method and system for providing a transaction platform for pre-owned merchandise
US10248954B2 (en) 2014-08-14 2019-04-02 Alibaba Group Holding Limited Method and system for verifying user identity using card features
US10755345B2 (en) 2014-12-03 2020-08-25 Alibaba Group Holding Limited System and method for secure account transfer
US10579973B2 (en) 2015-01-19 2020-03-03 Alibaba Group Holding Limited System for efficient processing of transaction requests related to an account in a database
US10249013B2 (en) 2015-02-03 2019-04-02 Alibaba Group Holding Limited Method and system for wireless payment of public transport fare
US9672462B2 (en) 2015-03-03 2017-06-06 Umm Al-Qura University Smart surface-mounted hybrid sensor system, method, and apparatus for counting
US10296636B2 (en) 2015-10-09 2019-05-21 Alibaba Group Holding Limited Efficient navigation category management
GB2545003B (en) * 2015-12-03 2019-05-01 Int Consolidated Airlines Group Queue monitoring based on imprint profiles
EP3179453A1 (en) * 2015-12-03 2017-06-14 International Consolidated Airlines Group Queue monitoring based on imprint profiles
GB2545003A (en) * 2015-12-03 2017-06-07 Int Cons Airlines Group Queue monitoring based on imprint profiles
US10268166B2 (en) 2016-09-15 2019-04-23 Otis Elevator Company Intelligent surface systems for building solutions
JP2019061318A (en) * 2017-09-25 2019-04-18 東洋インキScホールディングス株式会社 Moving direction specification method and pressure-sensitive sensor device
US11538039B2 (en) 2018-02-12 2022-12-27 Advanced New Technologies Co., Ltd. Method and system for facilitating risk control of an online financial platform
US11816714B2 (en) 2018-03-19 2023-11-14 Advanced New Technologies Co., Ltd. Service verification method and apparatus
CN109323785A (en) * 2018-11-28 2019-02-12 浙江科技学院 Subway carriage crowding shifts to an earlier date predictor and its forecasting procedure
CN109323785B (en) * 2018-11-28 2023-08-18 浙江科技学院 Advanced prediction device and prediction method for congestion degree of subway carriage
CN112907785A (en) * 2021-01-15 2021-06-04 广东趣买票科技有限公司 Scenic spot number control system and control device based on Internet of things
DE102021122969A1 (en) 2021-09-06 2023-03-09 Incoretex GmbH Method for the continuous monitoring of an occupancy of a passenger transport vehicle
EP4151496A1 (en) 2021-09-06 2023-03-22 Incoretex GmbH Method for continuously monitoring the occupancy of a passenger transport vehicle and passenger transport vehicle

Also Published As

Publication number Publication date
ES2151844B1 (en) 2001-07-16
EP0992949A3 (en) 2001-05-16
ES2151844A1 (en) 2001-01-01

Similar Documents

Publication Publication Date Title
EP0992949A2 (en) People counting system by pressure images
US7702132B2 (en) Systems and methods for determining if objects are in a queue
US5926518A (en) Device for measuring the number of pass persons and a management system employing same
US7903841B2 (en) Miniature autonomous agents for scene interpretation
US7382267B2 (en) Systems and methods for area activity monitoring and personnel identification
US7929730B2 (en) Method and system for object detection and tracking
US20030053659A1 (en) Moving object assessment system and method
US20030053658A1 (en) Surveillance system and methods regarding same
US20030123703A1 (en) Method for monitoring a moving object and system regarding same
WO2012029518A1 (en) Behavior analysis device
US20060274917A1 (en) Image processing techniques for a video based traffic monitoring system and methods therefor
CN107662872A (en) The monitoring system and its monitoring method of passenger conveyor
Sun et al. Vehicle reidentification using multidetector fusion
EP1560161A2 (en) Method and system for searching for events in video surveillance
EP1686544B1 (en) Device for determining the number of objects in a supervised area
KR20220074335A (en) Apparatus and Method for Object Counting by Type Using Artificial Intelligence-Based Edge Computing
KR100900494B1 (en) System for movement tracing of moving object and service method thereof
CN107221056A (en) The method stopped based on human bioequivalence
CN114092858A (en) AI-based community fire fighting access occupation detection and identification method
Cavallaro et al. Suggesting Just Enough (Un) Crowded Routes and Destinations.
CN107221058A (en) Intelligent channel barrier system
JP2005215909A (en) Town traffic state investigation system using moving picture processing technique
CN115410113A (en) Fall detection method and device based on computer vision and storage medium
Kerridge et al. Monitoring the movement of pedestrians using low-cost infrared detectors: initial findings.
CN114418310A (en) Method and system for detecting operation efficiency of single mechanical body construction based on attitude identification

Legal Events

Date Code Title Description
PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

AK Designated contracting states

Kind code of ref document: A2

Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE

AX Request for extension of the european patent

Free format text: AL;LT;LV;MK;RO;SI

RIN1 Information on inventor provided before grant (corrected)

Inventor name: LEON SANCHEZ, FRANCISCO

Inventor name: PALOMO LOPEZ, CARLOS

Inventor name: GARCIA BAYO, RAMON

Inventor name: LOPEZ MUDARRA,JUAN IGNACIO

17P Request for examination filed

Effective date: 20000615

PUAL Search report despatched

Free format text: ORIGINAL CODE: 0009013

AK Designated contracting states

Kind code of ref document: A3

Designated state(s): AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE

AX Request for extension of the european patent

Free format text: AL;LT;LV;MK;RO;SI

AKX Designation fees paid

Free format text: AT BE CH CY DE DK ES FI FR GB GR IE IT LI LU MC NL PT SE

17Q First examination report despatched

Effective date: 20040920

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: NUCLEO DE COMUNICACIONES Y CONTROL, S.L.

RAP1 Party data changed (applicant data changed or rights of an application transferred)

Owner name: ELIOP SEINALIA, S.L.

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE APPLICATION IS DEEMED TO BE WITHDRAWN

18D Application deemed to be withdrawn

Effective date: 20110714