JP5920886B2 - Server, system, program and method for estimating POI based on terminal position / orientation information - Google Patents

Description

  The present invention relates to a technique for processing geospatial information.

  Services that provide users of information terminals with various geospatial information, such as information about specific locations around the user's current location, advertisements, and the like are known. Particularly, in recent years, augmented reality technology (AR (Augmented Reality) technology) is used to superimpose virtual annotation metadata (annotation) on real space video to provide related information and advertisements. Service is drawing attention.

  As an example of a service using such AR technology, Non-Patent Document 1, for example, discloses a Sekai Camera (registered trademark) provided by Tonchi Dot Co., Ltd. According to the Sekai camera, an annotation called an air tag, which is related information of this place, is superimposed and displayed on a specific place in a real space video imaged by a user using an information terminal such as a smartphone.

  Here, when presenting information / advertisement using AR technology, it is important for the information provider / advertiser to accurately grasp the place where the user is interested, the so-called POI (Point Of Interest). . As a technique for grasping this POI, for example, there is LBS (Location-Based Service). In LBS, POI is grasped by grasping how many users exist in which area in the geospace, and specifying a popular area.

  For example, Patent Literature 1 and Non-Patent Literature 2 disclose a technique in which a POI is appropriately displayed on the screen of a mobile terminal based on the current position of the mobile terminal. In particular, the technique of Patent Document 1 specifies the yaw angle and pitch angle of a mobile terminal using an attitude sensor, and appropriately displays the POI on the screen in consideration of the current attitude of the mobile terminal.

JP 2011-022112 A

Tonchi Dot Co., Ltd., “Sekai Camera Support Center”, [online], [searched July 9, 2012], Internet <http://support.sekaicamera.com/how-to> IPSJ Journal, “Special Augmented Reality (AR)”, Vol.51, No.4, pp. 385-391, 2010

  However, even if the conventional techniques as described above are used, it is very difficult to grasp the object / point that the user actually sees with interest as POI.

  Actually, when presenting information / advertisement using AR technology, it is important to identify what object / point is a POI. Specifically, it is preferable that an object / point in a position / orientation visually recognized by many (high percentage) users is set as POI. Here, “visual recognition” includes visual recognition through an image taken by the camera of the information terminal.

  On the other hand, for example, in the conventional LBS, it is possible to specify a popular area where users are gathered. However, in the vicinity of the area, it is impossible to grasp what number of users / objects / points are actually viewed from which direction. Further, in Patent Document 1, the POI can be displayed not only for the current position of the information terminal but also for the current posture. However, this technology is based on the premise that there is a POI to which latitude / longitude information is added, and does not grasp a POI that can be visually recognized by many users.

  Furthermore, when presenting information / advertisement using AR technology, ascertaining from which POI to which the information / advertisement is given, what number (ratio) of the POI is viewed by which user from what direction Is also very important. As a result, information / advertisement can be selectively given to positions and orientations where the rate of visual recognition (degree of interest) is high. As a result, effective annotations are displayed at appropriate locations and the number of annotations is controlled, so that the user can easily reach the target information. However, even if the conventional techniques as described above are used, it is very difficult to obtain information such as a high visibility and the degree of interest of the user.

  Therefore, an object of the present invention is to provide a POI estimation server, system, program, and method that can estimate a POI that a user sees with interest.

According to the present invention, a POI estimation server that communicates with a plurality of information terminals via a network,
Position and direction information storage means for storing the position information and direction information of the information terminal acquired from the information terminal for each information terminal;
The information terminal group in which the position information and the direction information are individually associated is divided into a plurality of clusters based on the accumulated position information, and for each cluster, a cluster barycentric position that is an average of the position information in the cluster, Cluster information generating means for generating a cluster interest vector corresponding to the sum of the orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the maximum vector length cluster interest vector to the cluster interest vector direction of the first cluster, and a cluster interest vector having the second largest vector length There is provided a POI estimation server having POI determining means for determining an intersection point with a second half line extending from the cluster centroid position of the second cluster in the direction of the cluster interest vector of the second cluster to POI.

As one embodiment of the POI estimation server according to the present invention, the POI determining means includes:
When n is an odd number starting from 1, and the POI is determined using the nth cluster and the (n + 1) th cluster, at least (n + 2) among the cluster interest vectors of the remaining clusters not yet involved in the POI determination ) Subtract the projection component in the direction connecting the cluster centroid position of the (n + 2) th cluster and the determined POI from the cluster interest vector of the (n + 2) th cluster having the cluster interest vector of the second largest vector length. Update the vector as the new cluster interest vector of the (n + 2) th cluster,
Among the remaining clusters excluding the first cluster to the (n + 1) th cluster, the (n + 2) th cluster from the cluster centroid position of the (n + 2) th cluster having the maximum vector length cluster interest vector The cluster interest vector of the (n + 3) -th cluster from the cluster centroid position of the (n + 3) -th cluster having the (n + 2) -th straight line extending in the direction of the interest vector and the cluster interest vector of the second largest vector length. It is also preferable to determine the intersection point with the (n + 3) half line extending in the direction as the (n + 3) / 2nd POI.

  In the embodiment in which the projection component is subtracted and the cluster interest vector is updated, the POI determination unit is still involved in the POI determination when the POI is determined using the nth cluster and the (n + 1) th cluster. A vector obtained by subtracting the projection component in the direction connecting the cluster centroid position of each cluster and the determined POI from the cluster interest vector of each remaining cluster is updated as a new cluster interest vector for each cluster. It is also preferable.

  Furthermore, as another embodiment of the POI estimation server according to the present invention, the first cluster is included in the first azimuth range of the POI including the azimuth indicating the cluster centroid position of the first cluster as seen from the determined POI. The degree of interest corresponding to the vector length of the cluster interest vector of the second cluster is given, and the vector length of the cluster interest vector of the second cluster is set to the second orientation range of the POI including the orientation indicating the cluster centroid position of the second cluster. It is also preferable to further include an interest degree giving means for giving a corresponding degree of interest.

Here, the interest degree assigning means described above is configured such that when n is an odd number starting from 1, and the POI is determined using the nth cluster and the (n + 1) th cluster, the (n + 2) th largest vector length cluster interest In the cluster interest vector of the (n + 2) -th cluster having a vector, the degree of interest corresponding to the size of the projection component in the direction connecting the cluster centroid position of the (n + 2) -th cluster and the determined POI is calculated,
It is also preferable to give the degree of interest corresponding to the size of the projection component to the azimuth range of the POI including the azimuth indicating the cluster centroid position of the (n + 2) th cluster when viewed from the determined POI.

Furthermore, as another embodiment of the POI estimation server according to the present invention, an annotation accumulating unit that accumulates an annotation displayed in association with the POI for each azimuth range in the determined POI,
It is preferable to further include annotation search means for searching for an annotation to be superimposed and displayed on the real space image displayed on the screen of the information terminal using the position information and the direction information received from the information terminal as keys.

According to the present invention, the system further includes the POI estimation server described above and an information terminal that communicates with the POI estimation server via a network,
Information terminal
A positioning unit that measures the position of the information terminal itself;
An orientation sensor that measures the orientation of the information terminal itself,
There is provided a system characterized by comprising position / orientation information generation means for generating position information and direction information to be transmitted to the POI estimation server from the measured position and direction.

According to the present invention, there is further provided a POI estimation program mounted on a POI estimation server that communicates with a plurality of information terminals via a network,
Position and direction information storage means for storing the position information and direction information of the information terminal acquired from the information terminal for each information terminal;
The information terminal group in which the position information and the direction information are individually associated is divided into a plurality of clusters based on the accumulated position information, and for each cluster, a cluster barycentric position that is an average of the position information in the cluster, Cluster information generating means for generating a cluster interest vector corresponding to the sum of the orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the maximum vector length cluster interest vector to the cluster interest vector direction of the first cluster, and a cluster interest vector having the second largest vector length Provided is a POI estimation program for causing a computer to function as POI determining means for determining an intersection point with a second half line extending in the direction of the cluster interest vector of the second cluster from the cluster centroid position of the second cluster as POI. The

According to the present invention, there is further provided a POI estimation method based on software information processing in a computer having a server for POI estimation that communicates with a plurality of information terminals via a network,
A first step of storing, for each information terminal, position information and orientation information of the information terminal acquired from the information terminal;
The information terminal group in which the position information and the direction information are individually associated is divided into a plurality of clusters based on the accumulated position information, and for each cluster, a cluster barycentric position that is an average of the position information in the cluster, A second step of generating a cluster interest vector corresponding to the sum of orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the maximum vector length cluster interest vector to the cluster interest vector direction of the first cluster, and a cluster interest vector having the second largest vector length There is provided a POI estimation method comprising a third step of determining an intersection point with a second half line extending from the cluster centroid position of the second cluster in the direction of the cluster interest vector of the second cluster as the POI. .

  According to the POI estimation server, system, program, and method of the present invention, it is possible to estimate the POI that the user visually recognizes with interest.

It is a block diagram which shows roughly one Embodiment of the system comprised from the POI estimation server by this invention, and the information terminal connected via the network. It is the schematic which shows the procedure of the cluster information generation and POI determination which concern on this invention. It is the schematic which shows the procedure which determines 2nd POI based on this invention. It is the schematic which shows the procedure which provides the "interesting degree" for every direction to POI based on this invention. 3 is a flowchart schematically showing an embodiment of a POI estimation method and an “interesting degree” giving method according to the present invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings.

  FIG. 1 is a block diagram schematically showing an embodiment of a system including a POI estimation server according to the present invention and information terminals connected via a network.

  According to FIG. 1, the POI estimation server 1 is an information processing apparatus capable of communicating with a plurality of information terminals 2 via radio and the Internet. The POI estimation server 1 exists within a predetermined range of the map space (geographic space) with the following procedures (a) to (c) based on the position information and direction information of the information terminal 2 acquired from the information terminal 2. Estimate the POI.

(A) The location information and orientation information of the information terminal 2 acquired from the plurality of information terminals 2 are accumulated for each information terminal,
(B) The information terminal group in which the position information and the direction information are individually associated is divided into a plurality of clusters based on the accumulated position information. For each cluster, “cluster” is an average of the position information in the cluster. `` Center of gravity position '' and `` cluster interest vector '' corresponding to the sum of orientation information in the cluster,
(C) POI is determined based on “cluster interest vector” and “cluster centroid position” of the maximum vector length and “cluster interest vector” and “cluster centroid position” of the second largest vector length.

  On the other hand, the information terminal 2 is a device possessed by the user, such as a mobile phone, a smartphone, a tablet computer, a notebook computer, a PDA (Personal Digital Assistant), or a head mounted display (HMD). The information terminal 2 includes a communication interface 200, a positioning unit 201, an orientation sensor 202, a camera 203, a display 204, and a functional configuration unit as a processor memory. Here, the function configuration unit (processor memory) realizes its function by executing a program that causes the computer mounted on the information terminal 2 to function. This functional configuration unit includes a position / orientation information generation unit 210, an annotation acquisition unit 211, and a display control unit 212.

  Similarly, according to FIG. 1, the positioning unit 201 uses the GPS satellite 3, for example, and measures the location of the information terminal 2 by the GPS positioning method. In addition, the positioning unit 201 obtains latitude / longitude information of the base station (area) from a base station installed in the vicinity of the location at a location where a positioning signal from the GPS satellite 3 cannot be received, such as in a subway station. You may obtain it directly.

  The azimuth sensor 202 is, for example, a three-axis type acceleration sensor using a capacitance method or a piezoresistive method, and a magnetoresistive (AMR, GMR or TMR) effect, a magnetic impedance (MI) effect, a fluxgate (FG) method, or a hall And a triaxial type geomagnetic sensor that measures the geomagnetism using the effect.

  Among these, the acceleration sensor measures an angle formed by a predetermined reference axis of the information terminal 2 with respect to the direction of gravity acceleration, that is, the vertical direction. Further, the geomagnetic sensor measures an angle formed by a predetermined reference axis of the information terminal 2 with respect to the direction of the magnetic field lines. Thereby, the direction of the predetermined reference axis of the information terminal 2 is determined at least in the horizontal plane (in the map space). Here, if the predetermined reference axis is, for example, the optical axis of the camera 203, the direction of the camera 203, that is, the direction of the real space image displayed on the display 204 is determined by the direction sensor 202.

  It is preferable that the camera 203 is fixed and installed so that the shooting direction and the line-of-sight direction when the user of the information terminal 2 looks at the display 204 substantially coincide with each other. In this case, when the user directs the information terminal 2 to the target / point of interest and displays the video of the target / point on the display 204, the optical axis of the camera 203, that is, the (predetermined) of the information terminal 2 is displayed. The orientation of the reference axis is the orientation of the object / point of interest with reference to the information terminal 2.

  The position / orientation information generation unit 210 receives the measurement result from the positioning unit 201 and the measurement result from the direction sensor 202 and generates the position information and the direction information of the information terminal 2. The generated information is transmitted to the POI estimation server 1. Here, the location information can be, for example, the latitude / longitude of the location of the information terminal 2. Further, the azimuth information can be, for example, north as a reference azimuth, and an azimuth angle (in a horizontal plane) that is positive in the clockwise direction between the reference azimuth and the reference axis of the information terminal 2.

  The annotation acquisition unit 211 acquires an annotation from the POI estimation server 1. In addition, the annotation is output to the display control unit 212 in order to display the annotation in a superimposed manner (for example, as if it is floating) in the vicinity of the corresponding POI in the real space image displayed on the display 204, and the display is performed. Instruct.

  Similarly, according to FIG. 1, the POI estimation server 1 includes a communication interface 100 which is an interface on the POI estimation server 1 side when communicating with the information terminal 2 via radio and the Internet, and a functional configuration unit as a processor memory. And. Here, the function configuration unit (processor memory) realizes its function by executing a program that causes the computer mounted on the POI estimation server 1 to function. This functional configuration unit includes a position / orientation information storage unit 110, a cluster information generation unit 111, a POI determination unit 112, an interest level assignment unit 113, a POI / annotation storage unit 114, and an annotation search unit 115. ing.

  The position / orientation information storage unit 110 stores the position information and the direction information of the information terminal 2 acquired from the plurality of information terminals 2 for each information terminal. The position / orientation information storage unit 110 creates a position / orientation information table in which the latitude / longitude and the azimuth are described, for example, for each MAC (Media Access Control) address of the information terminal 2.

  The cluster information generation unit 111 includes a cluster generation unit 111a and an interest vector generation unit 111b. The cluster generation unit 111a distributes the information terminal group in which the position information and the direction information are individually associated with each other on the map space, and divides the information terminal group into a plurality of clusters based on the accumulated position information. The interest vector generation unit 111b generates, for each cluster, a “cluster centroid position” that is an average of the position information in the cluster and a “cluster interest vector” that corresponds to the sum of the orientation information in the cluster. These functions in the cluster information generation unit 111 will be described in detail later with reference to FIG.

  The POI determination unit 112 includes a first half line extending from the “cluster centroid position” of the first cluster having the “cluster interest vector” having the maximum vector length in the direction of the “cluster interest vector” of the first cluster. Intersection with the second half-line extending from the “cluster centroid position” of the second cluster having the “cluster interest vector” having the second largest vector length in the direction of the “cluster interest vector” of the second cluster Is determined as POI. This function in the POI determination unit 112 will also be described in detail later with reference to FIG.

  When the POI determination unit 112 further determines the POI using the first cluster and the second cluster, the “cluster interest vector” of the third cluster having the “cluster interest vector” having the third largest vector length. The vector obtained by subtracting the projection component in the direction connecting the “cluster centroid position” of the third cluster and the determined POI is updated as a new “cluster interest vector” of the third cluster. As will be described in detail later with reference to FIG. 3, for the fourth (and subsequent) clusters, a vector obtained by subtracting a similar projection component may be updated as a new “cluster interest vector”.

  Next, the POI determination unit 112 determines the third cluster from the “cluster centroid position” of the third cluster having the cluster interest vector of the maximum vector length among the remaining clusters excluding the first cluster and the second cluster. The fourth cluster from the “cluster centroid position” of the fourth cluster having the third half-line extending in the direction of the “cluster interest vector” of the cluster and the “cluster interest vector” of the second largest vector length The intersection with the fourth half line extending in the direction of “cluster interest vector” is determined as the second POI. These functions in the POI determination unit 112 will also be described in detail later with reference to FIG.

  The degree-of-interest giving unit 113 includes the first cluster in the first azimuth range of this POI including the azimuth indicating the “cluster centroid position” of the first cluster as viewed from the determined (first) POI. An “interest” corresponding to the vector length of the “cluster interest vector” is given, and the second cluster's “2” orientation range including the orientation pointing to the “cluster centroid position” of the second cluster is “ “Interest level” corresponding to the vector length of “cluster interest vector” is assigned.

  The interest degree assigning unit 113 also determines the “cluster interest vector” of the third cluster having the “cluster interest vector” having the third largest vector length when the POI is determined using the first cluster and the second cluster. "Interesting degree" corresponding to the size of the projection component in the direction connecting the "cluster centroid position" of the third cluster and the determined (first) POI is calculated. Next, the calculated “degree of interest” is given to the azimuth range of this POI including the azimuth indicating the “cluster centroid position” of the third cluster as seen from the determined (first) POI. These functions in the interest level assigning unit 113 will also be described in detail later with reference to FIG.

  The POI / annotation accumulating unit 114 accumulates the annotations displayed in association with the POI for each azimuth range in the determined POI. Here, the POI position information and the “interesting degree” given by the interest degree assigning unit 113 are also stored in association with the azimuth range of this POI. Further, the annotation search unit 115 searches for an annotation to be superimposed and displayed on the real space image displayed on the screen of the information terminal 2 using the position information and the direction information received from the information terminal 2 as keys. The retrieved annotation is transmitted from the communication interface 100 to the information terminal 2 via the Internet and wireless.

  FIG. 2 is a schematic diagram showing procedures for generating cluster information and determining POI according to the present invention.

  According to FIG. 2A, a plurality of dots each corresponding to one information terminal 2 are distributed in a predetermined range of map space (geographic space). Each of these dots includes position information (latitude / longitude) and azimuth information (azimuth angle) of the information terminal 2. This distribution chart shows information that the position / orientation information storage unit 110 (FIG. 1) acquires and stores from a plurality of information terminals 2, for example, positions where latitude / longitude and azimuth are described for each MAC address of the information terminal 2 Based on the azimuth information table, it is created by the cluster generation unit 111a (FIG. 1).

  Here, the range of the distribution map may be, for example, a range of one mesh when the map space is divided with a mesh of 100 m intervals. In the present invention, since the POI that is actually visually recognized by the user is estimated, it is preferable that the size of the mesh is also determined in consideration of the distribution state of the objects and points that can be visually recognized. Note that “visual recognition” includes visual recognition through an image taken by a camera of an information terminal.

  In addition, as the timing for acquiring the position information and direction information of the information terminal 2, from the information terminal 2 that has activated a predetermined application using AR within a certain period of time, or from the information terminal 2 that has further searched and collected some information The position information and the direction information may be acquired, for example, each time or periodically. In addition, when the user points the information terminal 2 in a certain direction with interest and maintains the direction within a predetermined time range, it is also preferable to acquire information on this direction together with position information at that time. Furthermore, when the direction of the information terminal 2 is fixed for a predetermined time or more, for example, when the information terminal 2 is facing a certain direction with respect to the traveling direction during the user's boarding, from such an information terminal 2, It is also preferable not to acquire position information and azimuth information, or it is nevertheless possible to acquire position information and azimuth information periodically.

  Next, as shown in FIG. 2B, the cluster generation unit 111a (FIG. 1) divides these dots (information terminal group) into a plurality of clusters based on the position information (latitude / longitude). This clustering can be performed using various known methods, but may be performed using, for example, a DBSCAN (Density-Based Spatial Clustering) method.

  The DBSCAN method is a clustering method that can extract clusters having an arbitrary shape. In the DBSCAN method, a direct density-reachable (DDR) connection relationship of objects (dots) is defined using two parameters, a distance threshold Eps and a target (dot) number threshold MinPts. Are classified as the same cluster.

Specifically, when the following conditions (1) and (2) are satisfied, the connection from x (dot) to y (dot) has a DDR relationship.
(1) y∈N _Eps (x)
(2) | N _Eps (x) | ≧ MinPts
Here, N _Eps (x) is a set of dots that are in the vicinity of x and whose distance from x is within the threshold value Eps. Also, | N _Eps (x) | is the number of dots included in the neighborhood N _Eps (x). Note that the distance between two dots in FIG. 2 (B) is expressed by using a and b as constants for distance conversion.
(A × (Latitude difference between dots) ² + b × (Latitude difference between dots) ² ) ^0.5
Is calculated as

  In the DBSCAN method, starting from a certain seed point (dot), dots having a DDR relationship satisfying the conditions (1) and (2) are sequentially traced while being classified into the same cluster, and finally reachable maximum A set is extracted as one cluster. Here, the local maximum means a state where a new DDR target cannot be found near the dots in the cluster. After one cluster is determined as a maximal set, dots in this cluster are removed, a new seed point (dot) is selected, and the next cluster is generated.

According to FIG. 2B, a plurality of dots (information terminal group) are divided into a plurality of clusters by the cluster generation unit 111a (FIG. 1). Next, the interest vector generation unit 111b (FIG. 1) generates “cluster centroid position” and “cluster interest vector” in the cluster for each cluster. Here, the “cluster centroid position” is an average of position information (for example, (latitude, longitude)) of dots in one cluster. For example, the latitude value of a dot (information terminal 2) i is LAT _i, and longitude If the value is LNG _i ,
(ΣLAT _i / n _dot , ΣLNG _i / n _dot )
Is generated as Here, Σ is a summation for i = 1, 2,..., N _dot (n _dot : the number of dots in the cluster).

The “cluster interest vector” is defined as a vector corresponding to the sum of orientation information in the cluster. Specifically, the azimuth information of the dot (information terminal 2) i in the cluster is regarded as the azimuth unit vector e _azi having the direction of the azimuth angle θ _azi of the information terminal 2, and all the azimuth unit vectors e in the cluster The vector sum of _azi is the “cluster interest vector” V _I , ie
(3) V _I = Σe _azi
And Here, Σ is a total sum of vector sums for i = 1, 2,..., N _dot .

In addition, the starting point of the "cluster interest vector" V _I, referred to as "cluster center of gravity" of this cluster. Thus, "Cluster interest vector" V _I is taken as an index representing the position and orientation information of the information terminal 2 in this cluster. Also, in the map space in a predetermined range dots (information terminal group) are distributed, the set V _set of all V _I is regarded as a vector set characterizing the map space.

Such a “cluster interest vector” V _I is obtained from the “cluster centroid position” when, for example, more information terminals 2 (azimuth unit vectors e _azi ) are directed to one object / point in the cluster. It has more components in the direction toward the object / point.

  In fact, for example, a user who has started an application that provides geospatial information using AR technology such as Sekai Camera (registered trademark), asks the subject / point of interest “What is there? In many cases, the information terminal 2 (the objective lens of the camera (the optical axis thereof)) is pointed.

Thus, the orientation of the "cluster interest vector" V _I is in generally face tendency by many people to which the user has directed the information terminal 2 with the interest of the object-point in the cluster. Further, as a result, the vector length of the V _I will correspond to the degree of interest of the user for such objects, the point (degree of interest).

In each cluster, the vector length | V _I | of the “cluster interest vector” V _I is compared with the sum Σ | e _azi | of the vector lengths of all orientation unit vectors e _azi in the cluster. 4) | V _I | <C × Σ | e _azi | (C: a predetermined coefficient _satisfying 0 <C <1)
If the above is true, it is also preferable to reset the range of the distribution map (map space) shown in FIG. 2 (A), assuming that the object / point of interest is scattered or dispersed. For example, it is also preferable to change the range of the distribution map from a mesh divided at 100 m intervals on the map space to a mesh range divided at 10 m intervals. In this way, for example, in the downtown area of the metropolis, because the degree of high object-point to be interested there are many, to reflect accurately the degree of interest of the user in order to obtain a "cluster interest vector" V _I It may also be necessary to limit the map space range to be investigated.

  Next, as shown in FIG. 2C, the POI determination unit 112 (FIG. 1) selects the cluster A as the first cluster having the “cluster interest vector” of the maximum vector length from the plurality of clusters. Then, the cluster B is selected as the second cluster having the “cluster interest vector” having the second largest vector length.

Next, the POI determination unit 112 performs a first half line extending from the “cluster centroid position” of the cluster A in the direction of the “cluster interest vector” V _IA of the cluster A, and the cluster B from the “cluster centroid position” of the cluster B. The intersection point with the second half-line extending in the direction of “cluster interest vector” V _IB is determined, and this intersection point is determined as the first POI.

Here, the “cluster interest vector” V _IA of the cluster A (first cluster) has the maximum vector length. This means that many users in the cluster A tend to point the information terminal 2 at an object / point existing in the direction of the “cluster interest vector” _VIA with interest. . The “cluster interest vector” V _{IB of} the cluster B (second cluster) also has a vector length next to V _IA . This also means that many users in the cluster B tend to point the information terminal 2 at an object / point existing in the direction of the “cluster interest vector” _VIB with interest. Yes.

  Therefore, the intersection of the first half-line of cluster A and the second half-line of cluster B described above is an object with a high degree of interest that many users in clusters A and B are interested in. It is understood that there is a high probability of corresponding to an object / point, that is, POI. If the intersection of the first and second half lines does not exist in the predetermined map space range, the POI estimation procedure may be terminated assuming that there is no POI with a high degree of interest.

As described above, according to the present invention, an information terminal group is clustered, and “cluster centroid position” and “cluster interest vector” V are used as indices representing the position / orientation information of the information terminals 2 in each cluster. _IA is calculated. Next, based on these amounts, an object / point with a high probability that many users are interested in pointing the information terminal 2 is determined as the POI. This makes it possible to estimate the POI that the user visually recognizes with interest.

  FIG. 3 is a schematic diagram showing a procedure for determining the second POI according to the present invention.

According to FIG. 3A, when the first POI is determined using the cluster A and the cluster B (FIG. 1C), the POI determination unit 112 further selects the “cluster” having the third largest vector length. Cluster C is selected as the third cluster having “interest vector”. Then, the “cluster interest vector” V _IC of cluster C is
(A) a projection component in a direction connecting the “cluster centroid position” and the determined first POI;
(B) It decomposes | disassembles into the non-projection component orthogonal to this projection component.

Then, POI determination unit 112 updates the non-projection component of the "Cluster interest vector" _{V IC,} a new "Cluster interest vector" cluster C (3-th cluster) _{V IC} '. In other words, from the "cluster interest vector" V _IC, a vector which is obtained by subtracting the projection component of V _IC, is updated as a new cluster C "Cluster interest vector" V _{IC '.}

  Thereafter, as shown in FIG. 3B, the POI determination unit 112 has the maximum vector length of the remaining clusters excluding the cluster A (first cluster) and the cluster B (second cluster). A third cluster having a “cluster interest vector” and a fourth cluster having a “cluster interest vector” having the second largest vector length are determined. In FIG. 3B, cluster C is selected as the third cluster, and cluster D is selected as the fourth cluster.

Next, the POI determination unit 112 determines from the “cluster centroid position” of the cluster C, the third half line extending in the direction of the “cluster interest vector” V _IC ′ of the cluster C, and the “cluster centroid position” of the cluster D. The intersection of the cluster D with the fourth half line extending in the direction of the “cluster interest vector” V _ID of the cluster D is determined as the second POI.

Here, as a change mode, as shown in FIG. 3C, when the first POI is determined using the cluster A and the cluster B, the POI determination unit 112 determines the remaining POIs that are not yet involved in the POI determination. The cluster interest vector of each cluster (in FIG. 3C, each of V _IC and V _ID ),
(A ′) a projection component in the direction connecting the “cluster centroid position” and the determined first POI;
(B ′) It is also preferable to decompose into a non-projection component orthogonal to the projection component.

In this case, POI determination unit 112 updates the non-projection component of the "Cluster interest vector" V _IC, a new "Cluster interest vector" cluster C (3-th cluster) V _{IC ',} further, "Cluster interest non projection component of the vector "V _ID, and updates as a new cluster D (4 th cluster)" cluster interest vector "V _{ID '.} In other words, a vector obtained by subtracting the projection component in the direction connecting the cluster centroid position of each cluster and the determined POI from the cluster interest vector of each remaining cluster that is not yet involved in POI determination is obtained for each cluster. As a new cluster interest vector.

  Further, the POI determination unit 112 subtracts the projection component from the cluster interest vectors of a predetermined number of clusters counted in descending order of vector length among the cluster interest vectors of the remaining clusters not yet involved in the POI determination. It is also possible to update the non-projection component as a new cluster interest vector. In any case, the amount of calculation for POI determination decreases as the number of cluster interest vectors to be updated decreases.

  Thereafter, returning to the step shown in FIG. 3A, the fifth cluster having the “cluster interest vector” of the maximum vector length is determined from among the clusters A to D, and this “cluster interest vector” is determined. Is updated as a new “cluster interest vector” of the fifth cluster. Next, the third POI is determined in the same manner as the step for determining the second POI described above.

In this way, a new POI can be determined sequentially by repeating the series of steps shown in FIG. Here, when the vector length of the non-projection component (new “cluster interest vector”) obtained in FIG. 3A (or FIG. 3C) is less than the predetermined threshold value l _th , the procedure for POI determination is as follows. It is also preferable to terminate. Thus, the fact that the vector length of the non-projection component (new “cluster interest vector”) is less than the predetermined threshold value l _th is interpreted that there is no object / point where the degree of interest is greater than or equal to the predetermined value.

  As described above, according to the embodiment shown in FIG. 3, after determining the (first) POI, the non-projection component in the “cluster interest vector” of the third cluster (or subsequent cluster) is determined. , A new “cluster interest vector”. In this way, by subtracting the degree of interest (projection component) with respect to the determined POI, it is possible to appropriately estimate the next new POI that the user can visually recognize with interest.

  FIG. 4 is a schematic diagram showing a procedure for assigning “degree of interest” for each direction to the POI according to the present invention.

In FIG. 4A, the first POI determined, the “cluster interest vectors” V _IA and V _{IB of} the clusters A and B related to this POI determination, and the second POI are calculated. The projected component V _ICp of the “cluster interest vector” V _IC of cluster C is shown. This projection component V _ICp is a vector component that is excluded as a component toward the first POI in the “cluster interest vector” V _IC when the second POI is determined. Hereinafter, considering this component as well, the “degree of interest” for the first POI is calculated.

The interest level assigning unit 113 (FIG. 1) _calculates “interests” corresponding to V _IA , V _IB, and V _ICp from the “cluster interest vectors” V _IA and V _IB , and the projection component V _{ICp of the} “cluster interest vector” V _IC. "Degree" is calculated. For example, as viewed from the first POI, the “degree of interest” related to the direction indicating the “cluster centroid position” of the cluster _A is Iv _A, and the “degree of interest” related to the direction indicating the “cluster centroid position” of the cluster B is Iv _B If the “degree of interest” relating to the orientation pointing to the “cluster centroid position” of cluster C is Iv _C , Iv _A , Iv _B and Iv _C are, for example,
(5) Iv _A = | V _IA | / (| V _IA | + | V _IB | + | V _ICp |)
(6) Iv _B = | V _IB | / (| V _IA | + | V _IB | + | V _ICp |)
(7) Iv _C = | V _ICp | / (| V _IA | + | V _IB | + | V _ICp |)
Is calculated as Here, | V _IA |, | V _IB |, and | V _ICp | are vector lengths of V _IA , V _IB, and V _ICp , respectively. In the definitions of the formulas (5) to (7), Iv _A + Iv _B + Iv _C = 1, and the “interest degree” is normalized to the maximum value 1 as a relative evaluation value.

Although Iv _A , Iv _B and Iv _C are not limited to the equations (5) to (7), the larger the values | V _IA |, | V _IB | and | V _ICp | It is preferable to take. That is, from the definition of “cluster interest vector”, the vector length preferably corresponds to “degree of interest”. Also, absolute evaluation values obtained by multiplying | V _IA |, | V _IB |, and | V _ICp | by a predetermined constant may be adopted as Iv _A , Iv _B, and Iv _C , respectively. It is also possible to calculate only Iv _A and Iv _{B using} only “cluster interest vectors” V _IA and V _IB related to POI determination.

FIG. 4B shows how “interesting” is given to POI when Iv _A = 0.45, Iv _B = 0.39, and Iv _C = 0.16. When viewed from the first POI, “interest” Iv _A = 0.45 is given to the azimuth range (first azimuth range) W _IA including the azimuth indicating the “cluster centroid position” of the cluster A, and the cluster B An azimuth range (second azimuth range) W _IB including an azimuth indicating the “cluster centroid position” of W is _assigned “interest” Iv _B = 0.39, and the azimuth indicating the “cluster centroid position” of cluster C is The degree of interest “Iv _C = 0.16” is given to the included azimuth range (third azimuth range) W _IC .

Here, it is preferable that the azimuth ranges W _IA , W _IB, and W _IC are each set to include azimuths indicating all the information terminals 2 in the clusters A, B, and C as viewed from the first POI. .

Further, the azimuth range W _NO is azimuth range that is not visible first POI is from the direction of this range. For example, when the first POI is a columnar advertising tower provided at a corner of a building, the advertising display wall of the advertising tower cannot be seen from behind the building. In this case, as viewed from the advertising pillar, the orientation range of the background is the W _NO.

When the first POI is also visually recognized from the information terminal 2 in the cluster D (FIG. 3B), the projection component V _IDp toward the first POI in the “cluster interest vector” V _ID of the cluster D is It is also possible to extract and apply the “degree of interest” corresponding to the vector length | V _IDp | to the azimuth range W _ID including the azimuth indicating the “cluster centroid position” of the cluster D.

  As described above, according to the present embodiment, with respect to the POI estimated based on the “cluster interest vector” of the cluster, for each POI azimuth range including the azimuth indicating the “cluster centroid position” as viewed from the POI. The “interesting degree” corresponding to the vector length of the “cluster interest vector” can be given. That is, not only the POI corresponding to the direction visually recognized by the user but also the degree of interest of the user can be estimated for each orientation viewed from the POI. Thereby, for example, annotations such as different related information and advertisements according to the given “degree of interest” can be displayed for each azimuth range. As described above, the ability to present different annotations depending on the orientation of the user as viewed from the POI (the direction in which the user views the POI) is one particular effect that can be achieved by applying the present embodiment to the AR technology. is there.

  Furthermore, after preparing a POI with a specified “interest” for each azimuth range, for example, each azimuth range of the POI is provided as an advertising placement medium with an advertising placement fee according to the given “interest” It is also possible to do. That is, by converting the degree of interest / popularity into numerical values, it is possible to set an advertising fee corresponding to the degree of interest / popularity for each azimuth range. It is also possible to present information / advertisement in an order corresponding to the “interest” or prioritizing a direction with a high “interest”.

  In addition, since information and advertisements can be selectively given to positions and orientations where the ratio (degree of interest) that is visually recognized in this way is high, for example, effective annotations are displayed at appropriate places and annotations are displayed. The number can be controlled. As a result, the user can easily reach the target information on the screen of the information terminal 2.

  Furthermore, according to the present embodiment, the POI and the “interesting degree” for each azimuth at the time / time zone / period when the position / azimuth information of the information terminal is collected are estimated. It is possible to appropriately grasp the presence of a POI that is not noticed and the appearance of a new POI. In addition, it is possible to grasp the temporal transition of POI, such as occurrence of POI, change of “interesting degree”, disappearance of POI, etc., for example, presentation of information / advertisement according to time zone and direction, time zone and It is also possible to set an advertisement fee for each direction.

  FIG. 5 is a flowchart schematically showing an embodiment of a POI estimation method and an “interesting degree” giving method according to the present invention.

(S500) The position information and direction information of the information terminal 2 are stored for each information terminal 2.
(S501) The information terminal group is divided (clustered) into a plurality of clusters based on the position information.
(S502) For each cluster, a "cluster centroid position" and a "cluster interest vector" are generated.

(S503) The POI determination loop (steps S503 to S509) is started. Here, n is an odd number (= 1, 3, 5,...), And the initial value of n is 1. In addition, a predetermined vector length threshold is set to l _th .
(S504) “Cluster centroid position” in each of the nth cluster having the “cluster interest vector” having the maximum vector length and the (n + 1) th cluster having the “cluster interest vector” having the second largest vector length And the intersection of the nth half line and the (n + 1) th half line defined from the “cluster interest vector” is determined as the POI.

(S505) A vector (non-projection component) obtained by subtracting the projection component from the “cluster interest vector” of the (n + 2) -th cluster having the “cluster interest vector” of the (n + 2) -th largest vector length is the (n + 2) -th As a new “cluster interest vector” for the cluster.
(S506) Seen from the determined POI, for each azimuth range including the azimuth indicating the “cluster centroid position” of the nth cluster and the (n + 1) th cluster, and the “cluster centroid position of the (n + 2) th cluster A “cluster interest vector” or “interest degree” corresponding to the vector length of the projection component is assigned to an azimuth range including an azimuth indicating “”.

(S507) Among the remaining clusters excluding the first cluster to the (n + 1) th cluster, the (n + 2) th cluster having the “cluster interest vector” of the maximum vector length and the second largest vector length (N + 3) th cluster having the “cluster interest vector”.
(S508) Increment n by 2 (n is the next largest odd value).

(S509) If the vector length of the “cluster interest vector” in the nth cluster is equal to or greater than a predetermined vector length threshold l _th and n satisfies the loop condition that is smaller than the number of generated clusters, Returning to S503, the POI determination loop is repeated (the next POI is determined). On the other hand, if the above loop condition is not satisfied, this POI estimation method is completed assuming that there are no more determinable POIs.

  For the various embodiments of the present invention described above, various changes, modifications, and omissions within the scope of the technical idea and the viewpoint of the present invention can be easily made by those skilled in the art. The above description is merely an example, and is not intended to be restrictive. The invention is limited only as defined in the following claims and the equivalents thereto.

DESCRIPTION OF SYMBOLS 1 POI estimation server 100, 200 Communication interface 110 Location direction information storage part 111 Cluster information generation part 111a Cluster generation part 111b Interest vector generation part 112 POI determination part 113 Interest degree provision part 114 POI and annotation storage part 115 Annotation search part 2 Information Terminal 201 Positioning unit 202 Direction sensor 203 Camera 204 Display 210 Position / orientation information generation unit 211 Annotation acquisition unit 212 Display control unit 3 GPS satellite

Claims (9)

A POI (Point Of Interest) estimation server that communicates with a plurality of information terminals via a network,
Position and direction information storage means for storing the position information and direction information of the information terminal acquired from the information terminal for each information terminal;
An information terminal group in which the position information and orientation information are individually associated is divided into a plurality of clusters based on the accumulated position information, and a cluster centroid that is an average of the position information in the cluster for each cluster. Cluster information generating means for generating a position and a cluster interest vector corresponding to the sum of orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the cluster interest vector of the maximum vector length in the direction of the cluster interest vector of the first cluster, and a cluster interest vector of the second largest vector length POI determining means for determining a point of intersection with a second half line extending from the cluster centroid position of the second cluster to the direction of the cluster interest vector of the second cluster as POI Estimated server. The POI determining means is
When n is an odd number starting from 1, and the POI is determined using the nth cluster and the (n + 1) th cluster, at least (n + 2) among the cluster interest vectors of the remaining clusters not yet involved in the POI determination ) Subtract the projection component in the direction connecting the cluster centroid position of the (n + 2) th cluster and the determined POI from the cluster interest vector of the (n + 2) th cluster having the cluster interest vector of the second largest vector length. Update the vector as a new cluster interest vector of the (n + 2) -th cluster,
Among the remaining clusters excluding the first cluster to the (n + 1) th cluster, the (n + 2) th cluster is determined from the cluster centroid position of the (n + 2) th cluster having the maximum vector length cluster interest vector. Cluster interest of the (n + 3) -th cluster from the cluster centroid position of the (n + 3) -th cluster having the (n + 2) -th half straight line extending in the direction of the cluster interest vector and the second largest vector-length cluster interest vector 2. The POI estimation server according to claim 1, wherein the intersection point with the (n + 3) half line extending in the vector direction is determined to be the (n + 3) / 2th POI.   When the POI is determined using the n-th cluster and the (n + 1) -th cluster, the POI determination means determines the cluster centroid of each cluster from the cluster interest vector of each remaining cluster that is not yet involved in the POI determination. The POI estimation server according to claim 2, wherein a vector obtained by subtracting a projection component in a direction connecting the position and the determined POI is updated as a new cluster interest vector of each cluster.   When viewed from the determined POI, the degree of interest corresponding to the vector length of the cluster interest vector of the first cluster is added to the first azimuth range of the POI including the azimuth indicating the cluster centroid position of the first cluster. And assigning an interest degree corresponding to the vector length of the cluster interest vector of the second cluster to the second azimuth range of the POI including the azimuth indicating the cluster centroid position of the second cluster. The POI estimation server according to claim 1, further comprising means. When the POI is determined using the nth cluster and the (n + 1) th cluster, where n is an odd number starting from 1, the interest degree assigning means has a cluster interest vector having the (n + 2) th largest vector length ( In the cluster interest vector of the (n + 2) th cluster, the degree of interest corresponding to the size of the projection component in the direction connecting the cluster centroid position of the (n + 2) th cluster and the determined POI is calculated,
The degree of interest corresponding to the size of the projection component is given to the azimuth range of the POI including the azimuth indicating the cluster centroid position of the (n + 2) -th cluster when viewed from the determined POI. The POI estimation server according to claim 4. Annotation storage means for storing the annotation displayed in association with the POI for each orientation range in the determined POI;
The apparatus further comprises annotation search means for searching for an annotation to be superimposed and displayed on a real space image displayed on the screen of the information terminal, using the position information and direction information received from the information terminal as keys. 4. The POI estimation server according to 4 or 5. A system comprising the POI estimation server according to any one of claims 1 to 6, and an information terminal that communicates with the POI estimation server via a network,
The information terminal
A positioning unit that measures the position of the information terminal itself;
An orientation sensor that measures the orientation of the information terminal itself,
A system comprising position and direction information generating means for generating position information and direction information to be transmitted to the POI estimation server from the measured position and direction. A POI estimation program installed in a server for POI estimation that communicates with a plurality of information terminals via a network,
Position and direction information storage means for storing the position information and direction information of the information terminal acquired from the information terminal for each information terminal;
An information terminal group in which the position information and orientation information are individually associated is divided into a plurality of clusters based on the accumulated position information, and a cluster centroid that is an average of the position information in the cluster for each cluster. Cluster information generating means for generating a position and a cluster interest vector corresponding to the sum of orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the cluster interest vector of the maximum vector length in the direction of the cluster interest vector of the first cluster, and a cluster interest vector of the second largest vector length A computer functioning as POI determining means for determining an intersection point with a second half line extending from the cluster centroid position of the second cluster having the second cluster in the direction of the cluster interest vector of the second cluster. POI estimation program. A POI estimation method by software information processing in a computer having a server for POI estimation that communicates with a plurality of information terminals via a network,
A first step of storing, for each information terminal, the position information and direction information of the information terminal acquired from the information terminal;
An information terminal group in which the position information and orientation information are individually associated is divided into a plurality of clusters based on the accumulated position information, and a cluster centroid that is an average of the position information in the cluster for each cluster. A second step of generating a position and a cluster interest vector corresponding to the sum of orientation information in the cluster;
A first half line extending from the cluster centroid position of the first cluster having the cluster interest vector of the maximum vector length in the direction of the cluster interest vector of the first cluster, and a cluster interest vector of the second largest vector length And a third step of determining an intersection point with a second half line extending from the cluster centroid position of the second cluster having the second cluster in the direction of the cluster interest vector of the second cluster. POI estimation method.

Images