JP6713953B2 - Shape estimating device, shape estimating method and program - Google Patents

Description

The present invention relates to a shape estimation device, a shape estimation method and a program.

A technique is known in which a shape of an imaged object is grasped by imaging the object with a camera device. Hereinafter, a target object whose shape is to be grasped is referred to as a “target object”.

However, the satellite image or the like may not be able to effectively grasp the shape of the target object when there is a shield or the target object is small. In addition, a normal fixed camera may not be able to capture an image of the entire object when the object is huge. Further, direct imaging of the object may not be preferable from the viewpoint of privacy and the like.

On the other hand, there is known a technique of estimating an object by using randomly distributed binary sensors (Non-Patent Document 1), but it is possible to estimate an object by using a convex binary sensor. It is known that there are only the area and the outer circumference (Non-Patent Document 2). The binary sensor is a sensor that detects whether or not there is an object within the sensing range.

Therefore, in recent years, a method of estimating the shape of an object using a composite sensor that is a combination of binary sensors has been studied.

H. Saito, K. Shimogawa, S. Shioda, and J. Harada, "Shape Estimation Using Networked Binary Sensors," Proc. IEEE INFOCOM, 2009. Hiroshi Saito, Local Information, Observable Parameters, and Global View, IEICE Trans. Communications, E96-B, 12, pp.3017-3027, 2013.

However, the method using the above-mentioned composite sensor only allows the estimation of the angles of the vertices of the object and the selection from simple figures, and cannot even estimate the shape. It is also difficult to reduce the size of the composite sensor and to arrange a large number of composite sensors at random.

Here, for example, a directional distance sensor such as an ultrasonic sensor is commercially available. Some of such directional distance sensors are small and easy to arrange randomly.

The embodiment of the present invention has been made in view of the above points, and an object thereof is to realize shape estimation of an object using a directional distance sensor.

In order to solve the above problems, a plurality of directional distance sensors installed in a predetermined area, an acquisition unit that acquires a sensing result of sensing an object that moves linearly at a predetermined speed in the area, respectively, Based on the characteristic parameter extracted by the extracting means for extracting the characteristic parameter of the first section included in the sensing result acquired by the acquiring means, and the characteristic parameter extracted by the extracting means, It is characterized in that it has an estimating means for estimating a shape parameter of a second section corresponding to one section, and a connecting means for estimating a connection relationship between the shape parameters estimated by the estimating means.

Shape estimation of an object can be realized using a directional distance sensor.

It is a figure which shows an example of the whole structure of the shape estimation system in embodiment of this invention. It is a figure explaining an example of sensing by the shape presumption system in an embodiment of the invention. It is a figure which shows an example of the hardware constitutions of the shape estimation apparatus in embodiment of this invention. It is a figure which shows an example of a functional structure of the shape estimation apparatus in embodiment of this invention. It is a flow chart which shows an example of processing which a shape estimating device in an embodiment of the invention performs. It is a figure explaining an example of a sensing result. It is a figure explaining an example of the effect of the present invention.

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

<Overall structure>
First, the overall configuration of the shape estimation system 1 according to the embodiment of the present invention will be described with reference to FIG. FIG. 1 is a diagram showing an example of the overall configuration of a shape estimation system 1 according to an embodiment of the present invention.

The shape estimation system 1 shown in FIG. 1 includes a shape estimation device 10 and a plurality of sensors 20. The shape estimation device 10 and the plurality of sensors 20 are communicably connected via, for example, a wireless communication network.

The shape estimation device 10 is a computer that estimates the shape of an object using the result sensed by the sensor 20 (hereinafter referred to as “sensing result”). The shape estimation device 10 includes a shape estimation program 100 and a sensing result storage unit 200.

The shape estimation device 10 causes the sensing result storage unit 200 to store the sensing result received from the sensor 20 by the shape estimation program 100. In addition, the shape estimation device 10 uses the shape estimation program 100 to perform a process of estimating the shape of an object using the sensing result stored in the sensing result storage unit 200.

The sensor 20 is, for example, a directional distance sensor such as an ultrasonic sensor. The sensor 20 senses the distance to the target object at predetermined time intervals, and transmits the sensing result to the shape estimation device 10. Note that, hereinafter, when distinguishing each of the plurality of sensors 20, they are referred to as “sensor 20 ₁ ”, “sensor 20 ₂ ”, and the like.

The configuration of the shape estimation system 1 shown in FIG. 1 is an example, and other configurations may be used. For example, the shape estimation device 10 may be composed of a plurality of computers. Further, for example, the shape estimation program 100 may be composed of a plurality of modules. Furthermore, for example, the sensing result storage unit 200 may be included in another device connected to the shape estimation device 10 via a network.

<Sensing in the shape estimation system 1>
Here, sensing by the shape estimation system 1 according to the embodiment of the present invention will be described with reference to FIG. FIG. 2 is a diagram illustrating an example of sensing by the shape estimation system 1 according to the embodiment of the present invention.

The shape estimation system 1 in the embodiment of the present invention estimates the shape of the target T by sensing the target T moving in a predetermined monitoring area A. It is assumed that the object T enters the monitoring area A, moves along a linear trajectory at a constant speed v, and leaves. At this time, the shape, position, and moving speed v of the object T are unknown.

Further, the target T has a two-dimensional spread, and its contour is composed of line segments. Hereinafter, the moving direction of the object T will be referred to as a reference direction of the angle and the direction.

In the monitoring area A, n _s sensors 20 are randomly installed (that is, n _s sensors 20 are scattered). In the example shown in FIG. 2, the four sensors 20 of the sensor 20 ₁ , the sensor 20 ₂ , the sensor 20 _3, and the sensor 20 ₄ are randomly installed in the monitoring area A.

At this time, the maximum sensing range of each sensor 20 is r _max . Further, the position of each sensor 20 and the sensing area direction θ are unknown. Each sensor 20 senses the distance r to the object T in the sensing area direction θ at each time t, and transmits the sensing result r to the shape estimation device 10. The sensing area direction θ differs for each sensor 20.

When the position of the sensor 20 is included in the target object T, each sensor 20 transmits the sensing result r=0 to the shape estimation device 10. Further, each sensor 20 transmits the sensing result “no detection” to the shape estimation device 10 when the distance from the sensor 20 to the target T in the direction θ exceeds r _max .

Since the sensing result r changes with the change of the time t, the sensing result of the sensor 20 at the time t is represented as r(t). Further, the sensing result of the sensor 20 _i at time t is represented as r _i (t).

The area and perimeter of A in the monitoring area are sufficiently larger than the size of the object T and the maximum sensing range r _max .

As described above, in the shape estimation system 1 according to the embodiment of the present invention, the plurality of sensors 20 randomly arranged in the monitoring area A sense the object T moving on a linear trajectory at a constant speed v. To do. In the example shown in FIG. 2, the shape estimation device 10 is installed in the monitoring area A, but the shape estimation device 10 does not need to be installed in the monitoring area A.

<Hardware configuration>
Next, the hardware configuration of the shape estimation device 10 according to the embodiment of the present invention will be described with reference to FIG. FIG. 3 is a diagram showing an example of a hardware configuration of the shape estimation device 10 according to the embodiment of the present invention.

The shape estimation device 10 shown in FIG. 3 includes an input device 11, a display device 12, an external I/F 13, a RAM (Random Access Memory) 14, a ROM (Read Only Memory) 15, and a CPU (Central Processing Unit). 16, a communication I/F 17, and an auxiliary storage device 18. These pieces of hardware are connected to each other via a bus B so that they can communicate with each other.

The input device 11 is, for example, a keyboard, a mouse, a touch panel, etc., and is used by the user to input various operations. The display device 12 is, for example, a display, and displays the processing result of the shape estimation device 10.

The external I/F 13 is an interface with an external device. The external device includes a recording medium 13a and the like. The shape estimation device 10 can read or write the recording medium 13a or the like via the external I/F 13. The shape estimation program 100 and the like may be recorded in the recording medium 13a.

The recording medium 13a includes, for example, a flexible disk, a CD (Compact Disc), a DVD (Digital Versatile Disk), an SD (Secure Digital) memory card, a USB (Universal Serial Bus) memory card, and the like.

The RAM 14 is a volatile semiconductor memory that temporarily holds programs and data. The ROM 15 is a non-volatile semiconductor memory that can retain programs and data even when the power is turned off. The ROM 15 stores, for example, OS (Operation System) settings and network settings.

The CPU 16 is an arithmetic device that reads programs and data from the ROM 15 and the auxiliary storage device 18 onto the RAM 14 and executes processing.

The communication I/F 17 is an interface for connecting the shape estimation device 10 to a network. The shape estimation program 100 and the like may be acquired (downloaded) from a predetermined server or the like via the communication I/F 17.

The auxiliary storage device 18 is, for example, a HDD (Hard Disk Drive) or an SSD (Solid State Drive), and is a non-volatile storage device that stores programs and data. The programs and data stored in the auxiliary storage device 18 include, for example, an OS, an application program that implements various functions on the OS, a shape estimation program 100, and the like.

The shape estimating apparatus 10 according to the embodiment of the present invention can realize various processes described later by having the hardware configuration shown in FIG.

<Functional configuration>
Next, a functional configuration of the shape estimation device 10 according to the embodiment of the present invention will be described with reference to FIG. FIG. 4 is a diagram showing an example of a functional configuration of the shape estimation device 10 according to the embodiment of the present invention.

The shape estimation device 10 shown in FIG. 4 includes a sensing result management unit 101, a speed estimation unit 102, an extraction unit 103, a first shape estimation unit 104, a second shape estimation unit 105, and a connection relation estimation unit. It has 106, the correction|amendment part 107, and the output part 108. Each of these functional units is realized by the processing that the shape estimation program 100 causes the CPU 16 to execute.

Further, the shape estimation device 10 shown in FIG. 4 has a sensing result storage unit 200. The storage unit can be realized by using, for example, the auxiliary storage device 18.

The sensing result storage unit 200 stores the sensing result r(t) received from the sensor 20. The sensing result r(t) is stored in the sensing result storage unit 200 by the sensing result management unit 101. After that, in the sensing result storage unit 200, a plurality of objects T randomly installed in the monitoring area A from the time when the target T enters the monitoring area A to the time when the target T leaves (that is, passes). It is assumed that the sensing result r(t) of the sensor 20 is stored.

The sensing result management unit 101 acquires the sensing result r(t) stored in the sensing result storage unit 200. When the sensing result management unit 101 receives the sensing result r(t) from the sensor 20, the sensing result management unit 101 stores the sensing result r(t) in the sensing result storage unit 200.

The speed estimation unit 102 estimates the speed v of the object T. The reason why the velocity v of the object T is estimated is that the sensing result r(t) depends on the velocity v. Note that the velocity v can also be estimated as a by-product of shape estimation, but since the sensing result r(t) depends on the velocity v as described above, the velocity v itself is not a shape parameter, but it is necessary to estimate the velocity v. Become.

The extraction unit 103 extracts, from the sensing result r(t) of each sensor 20, the length l _d and the slope s _{d of} the line segment forming the sensing result r(t). A plurality of sets _{(l d, s} d) is extracted by the extraction unit 103. Incidentally, extracted by the extraction unit 103 _{(l d, s} d) is an example of a characteristic parameter.

The first shape estimation unit 104, a plurality of (l _{d, s d)} extracted by the extraction unit 103 by using, among the line segments constituting the contour of the object T, the line segment parallel to the moving direction presume. The estimated line segment is represented by a set of an estimated value of the length of the line segment and an estimated value of the direction of the line segment.

Here, since there are many objects T that have contour portions parallel to the moving direction and the line segments that form the contour portions parallel to the moving direction are easy to estimate, in the present embodiment, first, the moving direction Estimate the line segments that make up the contour portion parallel to.

The second shape estimation unit 105, a plurality of (l _{d, s d)} extracted by the extraction unit 103 by using, among the line segments constituting the contour of the object T, except a line segment parallel to the moving direction Estimate the line segment of. The estimated line segment is represented by a set of an estimated value of the length of the line segment and an estimated value of the direction of the line segment.

Here, the second shape estimation unit 105 has a provisional estimation unit 111, a consistency determination unit 112, and an adoption unit 113.

Preliminary estimation unit 111, two sets of _{(l d,} s _d), obtained from _{(l d ', s d'} ), the length and the provisional estimated value of the slope of the line segments constituting the contour of the object T.

Consistency determination unit 112 determines the provisional estimated value obtained by the provisional estimation unit _111, whether or not consistent with _{(l d'', s d''} ).

The adoption unit 113 adopts the provisional estimated value obtained by the provisional estimation unit 111 as an estimated value according to the determination result by the consistency determination unit 112. The estimated value adopted by the adoption unit 113 is a set of an estimated value of the length of a line segment other than a line segment parallel to the moving direction and an estimated value of the direction.

The pair of the length estimation value and the direction estimation value estimated by the first shape estimating unit 104 and the second shape estimating unit 105 is an example of the shape parameter.

The connection relation estimation unit 106 estimates the connection relation of the line segments estimated by the first shape estimation unit 104 and the second shape estimation unit 105 (that is, how the estimated line segments are connected). ..

When the target T has a concave portion, the correction unit 107 corrects the estimated omission of a line segment that forms a part of the contour of the target T based on the sensing result r(t) related to the concave portion. In addition, when the target T is convex and the estimation is performed well, the correction by the correction unit 107 is not performed.

The output unit 108 outputs the finally obtained estimation result (that is, the estimated line segment and the estimated line segment connection relationship). The output destination of the output unit 108 may be the display device 12 or the auxiliary storage device 18. Further, it may be another device or the like connected via the recording medium 13a or a network.

<Details of processing>
Next, details of the processing executed by the shape estimation apparatus 10 according to the embodiment of the present invention will be described with reference to FIG. FIG. 5 is a flowchart showing an example of processing executed by the shape estimation device 10 according to the embodiment of the present invention.

Step S1: The sensing result management unit 101 acquires the sensing result r(t) stored in the sensing result storage unit 200. For example, the sensing result management unit 101 detects the sensing result r ₁ (t) of the sensor 20 ₁ at each time t, the sensing result r ₂ (t) of the sensor 20 ₂ at each time t, and the sensing result of the sensor 20 ₃ at each time t. The result r ₃ (t) and the sensing result r ₄ (t) of the sensor 20 ₄ at each time t are acquired.

Step S2: The speed estimation unit 102 estimates the speed v of the object T. The velocity v of the object T can be estimated as follows.

Of n _s installed in the monitoring area A, the number of sensors 20 that detect the object T and never reach r(t)=0 is n _v . At this time, if the time for the object T is required to pass through the surveillance area A _{m t,} the area of the monitoring area A and _{A size, n v} expected value E _{[n v]} of the formula 1 below It is represented by.

ｍ_ｔは、監視領域Ａ内に設置されている全てのセンサ２０うちで、最も遅くまで対象物Ｔを検出していたセンサ２０のセンシング結果ｒの時刻から、最も早く対象物Ｔを検出したセンサ２０のセンシング結果ｒの時刻を引くことで算出することができる。

m _t is, among all of the sensors 20 installed in the surveillance area A, from the slowest to the sensing result r of the sensor 20 which has detected the object T time, was detected earliest object T sensor It can be calculated by subtracting the time of the sensing result r of 20.

On the other hand, the observed value of n _v is obtained from the number of actual sensors 20 that have detected the object T and have never reached r(t)=0 among the n _s sensors 20. Such an actual number of sensors 20 can be calculated from the sensing result r(t) acquired in step S1.

The observed values of n _v, instead of the expected value E [n _v], by using the left side of Equation 1 above, it is possible to obtain the value of the speed v (i.e., the estimated value of the velocity v).

Step S3: extracting section 103 extracts from the sensing result r (t) of each sensor 20 _{(l d, s} d). Set of the length _{l d} and inclination _{s d (l d, s d} ) may be as follows extracted from the sensing result r (t).

First, the sensing result r(t) of each sensor 20 will be described with reference to FIG. FIG. 6 is a diagram illustrating an example of the sensing result r(t).

As shown in FIG. 6, the object T moves along a linear trajectory in the monitoring area A at a constant velocity v, and its contour is composed of line segments. r(t) is composed of a combination of line segments.

For example, assume that the sensing area direction of the sensor 20 ₁ is θ ₁ . In this case, as in shown in FIG. 6, the sensor 20 _1, time t = _{t a} (relative position of the sensor 20 ₁ with respect to the object T is _x 1 _(t a)) when, for detecting a target T .. The sensor 20 ₁ during the time _t = t _a ~ time t = _{t b,} the distance between the object T _r 1 (t) is continuously sensing (detection). Thereafter, the sensor 20 _1, once past the time t = _{t b} (i.e., the relative position of the sensor 20 ₁ After past the _x 2 _{(t a)),} does not detect an object T. As a result, the sensing result r ₁ (t) of the sensor 20 ₁ is obtained.

Similarly, for the sensors 20 ₂ , 20 ₃ and 20 ₄ , the sensing result r ₂ (t), the sensing result r ₃ (t) and the sensing result r ₄ (t) are obtained according to the movement of the object T. To be

Here, each line segment forming the sensing result r(t) corresponds to at least a part of one line segment forming the contour of the object T. Then, in particular, among the line segments forming the sensing result r(t), the line segment satisfying the following conditions (1) to (3) corresponds to the entire line segment forming the contour of the object T. To do.

(1) In the line segment of the sensing result r(t), r(t)>0 is always satisfied.

(2) The line segment of the sensing result r(t) starts with a change in slope from the immediately preceding line segment section, starts with a discontinuous decrease, or immediately after the target T is not detected. _Is detected and a value smaller than r _max is detected.

(3) Immediately after the line segment of the sensing result r(t) ends with a change in the slope to the line segment section immediately after, ends with a discontinuous rise, or immediately after detecting a value smaller than r _max. The target T is no longer detected and the process ends.

Among the line segments forming the sensing result r(t), the line segment satisfying the above (1) to (3) (that is, the line segment corresponding to the entire line segment forming the contour of the object T). For each time, the length l _d and the slope s _{d of the} line segment are obtained. Thus, sensing result r (t) from _{(l d, s} d) of each sensor 20 is extracted.

Incidentally, one sensor 20 by detecting a plurality of line segments that constitute the contour of the object T, the set of sensing result of the sensor 20 r (t) from a plurality of (l _{d, s d)} is obtained. From the sensing result r (t), 1 single _{(l d, s} d) it is also likely may not be obtained. A plurality of (l _{d, s d)} obtained from the sensing result r (t) of the plurality of sensors 20 by using the shape estimation will be described later is performed.

Step S4: the first shape estimation unit 104, a plurality of (l _{d, s d)} extracted by the extraction unit 103 by using, among the line segments constituting the contour of the object T, parallel to the direction of movement Estimate the line segment. Of the line segments forming the contour of the object T, the line segments parallel to the movement direction can be estimated as follows.

Length and angle of the line segments constituting the contour of the target object T and λ respectively xi], if the sensing area direction of the sensor 20 is theta, l _d and s _d is the Equations 2 and 3 below, respectively Can be represented.

ここで、対象物Ｔの輪郭を構成する線分のうち、移動方向に平行な線分では、ξ＝０，πであるが、上記の式３から分かるように、ξ＝０，πと、ｓ_ｄ＝０とは等価である。また、上記の式２から、ξ＝０，πであれば、λ＝ｌ_ｄ×ｖである。

Here, among the line segments forming the contour of the object T, the line segments parallel to the movement direction have ξ=0,π, but as can be seen from the above formula 3, ξ=0,π, Equivalent to s _d =0. Further, from equation 2 above, xi] = 0, if [pi, is λ = _{l d} × v.

Therefore, if (l _d , s _d ) is obtained, if there is one that satisfies s _d =0, there is a line segment that constitutes a contour parallel to the moving direction, and the length of the line segment (that is, , The estimated value of the length λ of the line segment) is obtained as “ _ld ×v”. The angle of the line segment (that is, the estimated value of the angle of the line segment) is “0, π”. Note that the v uses the “estimated value of the speed v” estimated in step S2.

As a result, a set of the estimated value of the length of the line segment parallel to the moving direction and the estimated value of the direction is obtained among the line segments forming the contour of the object T.

Here, there may be a plurality of line segments forming a contour parallel to the moving direction. There may be a plurality of line segments having the same length, and there may be a plurality of line segments having different lengths. If different length of the line segment there are a plurality, as discussed _above, (l d, _{s d)} for each, the estimated value of the line segment length λ "l _d × v" is obtained. In this case, different estimates of (l _{d, s d)} length λ obtained from the "l _d × v", when there is close values, estimates of length λ indicating the same line Judge as a value. An existing clustering tool or the like may be used for such a determination.

Even if the clustering is performed, it is necessary to determine how many line segments having the same length among the line segments forming the contour parallel to the movement direction. First, it obtained when the length and angle of the line segments constituting the contour is lambda and xi], respectively (l _{d, s d)} the number of n _{d (λ,} ξ) and. At this time, based on the expected value of n _d (λ, ξ), it is estimated how many corresponding line segments (that is, line segments having the same length).

n d _(λ, ξ) expectation _{E [n d (λ, ξ} )] of, by analyzing the position of the sensor 20 that the length and angle can be observed across the line segment which is lambda and xi], respectively, the subject When the object T is convex, it is given by the following equations 4 and 5.

したがって、ｎ_ｄ（λ，ξ）の観測値がＥ［ｎ_ｄ（λ，ξ）］の約ｋ倍であれば、長さ及び角度がそれぞれλ及びξである線分がｋ本あると推定すれば良い。なお、この推定において、ｖ、λ及びξは、それぞれ上記の推定値（すなわち、「速度ｖの推定値」、「長さλの推定値」及び「角度ξの推定値」）を使用する。

Thus, n d _(λ, ξ) observations of _{E [n d (λ, ξ} )] about if k times, length and angle estimated that the line segment is lambda and xi] respectively is k book Just do it. In this estimation, v, λ, and ξ use the above-mentioned estimated values (that is, “estimated value of velocity v”, “estimated value of length λ”, and “estimated value of angle ξ”).

Step S5: the second shape estimation unit 105, a plurality of (l _{d, s d)} extracted by the extraction unit 103 by using, among the line segments constituting the contour of the object T, parallel to the direction of movement Estimate line segments other than line segments. Of the line segments forming the contour of the object T, line segments other than the line segments parallel to the moving direction can be estimated by the following steps S5-1 to S5-3.

Step S5-1: At first, the second shape estimation unit 105, the provisional estimation unit 111, two pairs of _{(l d,} s _d), the _{(l d ', s d'} ) from the contour of the target object T A tentative estimated value of the length λ and the slope ξ of the constituent line segment is obtained. These provisional estimates can be obtained as follows.

_{(L d,} s _d), with respect to _{(l d ', s d'} ), since the Equations 2 and 3 above is satisfied, λ, ξ, θ, solving simultaneous equations to unknowns θ' Thus, as λ, λ(+) shown in the following formula 6 and λ(−) shown in the formula 7 are obtained. Note that the v uses the “estimated value of the speed v” estimated in step S2.

また、λとして、λ（＋）とλ（−）とを用いて、以下の式８によりξ_０を算出する。なお、λ（＋）又はλ（−）が存在しない場合もある。この場合は、存在するもののみを使用してξ_０を算出すれば良い。

Further, using λ(+) and λ(−) as λ, ξ ₀ is calculated by the following equation 8. Note that λ(+) or λ(−) may not exist. In this case, only existing ones should be used to calculate ξ ₀ .

そして、以下の式９によりξを得る。

Then, ξ is obtained by the following Expression 9.

以上により得られたλ及びξが暫定推定値である。仮に、（ｌ_ｄ，ｓ_ｄ）、（ｌ_ｄ´，ｓ_ｄ´）が対象物Ｔの輪郭を構成する同一の線分をセンシングした結果得られたものであれば、上記で得られたλ及びξの暫定推定値は、真のλ及びξに近い値であることが期待できる。以降では、上記で得られた暫定推定値λを「λ_ａ」、暫定推定値ξを「ξ_ａ」と表す。

Λ and ξ obtained as described above are provisional estimated values. _{If, (l d, s d)} , (l d ', s d') if it has been obtained a result of the sensing of the same line segment that constitutes the contour of the object T, obtained above λ It can be expected that the provisional estimated values of ξ and ξ are close to true λ and ξ. Hereinafter, the provisional estimated value λ obtained above is represented as “λ _a ”, and the provisional estimated value ξ is represented as “ξ _a ”.

Step S5-2: Next, a second shape estimation unit 105, the consistency determination unit 112, the provisional estimated value obtained by the provisional estimation unit 111 (λ _a, ξ _a) is, _(l d'', s _{d ″} ). Provisional estimated value (λ _a, ξ _{a) is, (l d'', s d} '') whether consistent with, it can be determined as follows.

First, using l _{d ″} , s _{d ″} , the estimated value of the velocity v, and the provisional estimated value λ _a , ξ is obtained from the above equations 8 and 9. Then, the obtained ξ and the provisional estimated value ξ _a are compared, and if they are substantially the same, it is determined that they match, and if they are not the same, it is determined that they do not match (that is, conflict). Here, the fact that ξ obtained above is approximately the same as the provisional estimated value ξ _a means that the absolute value of the difference between these values is equal to or less than a preset positive constant C, for example. ..

Provisional estimated value obtained by the provisional estimation unit 111 (λ _a, ξ _a) each, all of _{(l d'', s d''} ) with, it is determined above. However, the provisional estimated value (λ _a, ξ _a) for each part of _{(l d'', s d''} ) using only, determination may be performed as described above.

Incidentally, finally adopted provisional estimated value as an estimated value (i.e., the provisional estimated value adopted by the adopted unit 113) _{and, (l d'', s d} '') if and are aligned, the estimated value _and referred to as estimated values that are consistent with _{(l d'', s d''} ).

Step S5-3: Finally, the second shape estimation unit 105 causes the adoption unit 113 to determine the provisional estimation value (λ _a , ξ) obtained by the provisional estimation unit 111 according to the determination result of the consistency determination unit 112. _{Take a} ) as the estimate.

Namely, the provisional estimated value (λ _a, ξ _a) of the matching to _{(l d'', s d''} ) in descending order of the provisional estimated value of a predetermined number (λ _a, ξ _a) estimate ( That is, it is adopted as a set of the estimated value of the length λ and the estimated value of the angle ξ. This, in the above step _S5-2, can be realized by storing the _{(l d'', s d''} ) for each, the provisional estimated value matching (λ _a, ξ _a).

As a result, a set of the estimated value of the length and the estimated value of the direction of the line segment other than the line segment parallel to the moving direction among the line segments forming the contour of the object T is obtained.

Besides determination method described in the above step S5-2, for example, by using the provisional estimates lambda _a, variable part of the arccos in Equation 8 above ^{_{(λ / v) 2 + l}} d 2 (1-s d 2) ) on which is calculated, and the calculation result, and Arccoskushi _a there is also a method of determining whether consistent with provisional estimates xi] _a.

Incidentally, in the above step _{S5-1 (l d, s d)} , (l d ', s d') , all of _{(l d, s} d) may be any two sets chosen from the set of, because it is a lot number of combinations for performing the above steps S5-1 in all sets, seems to have been obtained as a result of sensing the same line segment that constitutes the contour of the object T to (l _{d, s d)} It is efficient to select two sets in the same class as one class.

For example, it can be expected that efficiency is improved by classifying into the following classes according to the s _d value and selecting two sets in the same class. That is, a class where s _d is positive and the absolute value is very small, a class where s _d is negative and the absolute value is very small, a class where s _d is much larger than 1, a class where s _d is much smaller than −1, and s _d is A class that is positive and is not included in another class, and a class that s _d is negative and is not included in another class. Note that the absolute value is very small means that it is smaller than a certain positive constant ε that is set in advance. Also, being much larger than 1 (or much smaller than -1) means that s _d is 1+D or more (or s _d is -1 -D or less) with respect to a preset positive constant D. ..

For example, if s _d is positive and the absolute value is very small, then ξ that causes this s _d is slightly greater than 0 or slightly less than π. If there are one or two line segments with such xi], selected from the class _{(l d, s d),} (l d ', s d') can be a result of sensing the same line Becomes higher.

There may be a plurality of line segments that have the same length and whose angles are difficult to identify. In this case, as described in step S4, E[ _nd (λ, ξ)] is calculated by Equations 4 and 5, and the observed value of n _d (λ, ξ) is E[ _nd If it is approximately k times (λ, ξ)], it can be estimated that there are k line segments of λ and ξ having lengths and angles, respectively.

Step S6: Next, the connection relation estimation unit 106 determines the connection relation of the line segments estimated by the first shape estimation unit 104 and the second shape estimation unit 105 (that is, how the estimated line segments are connected. Estimate).

Here, among the plurality of sensors 20 installed in the monitoring area A, some of the sensors 20 sense all of a plurality of continuous line segments among the line segments forming the contour of the object T. There is. A continuous line segment is a line segment whose one ends are connected to each other.

Among the line segments forming the contour of the object T, two continuous line segments are generated in the sensing result r(t) of all the continuous line segments. Thus, two line segments successive in sensing result r _(t), can be obtained _{(l d, s d),} (l d ', s d').

On the other _{hand, (l d, s d)} , is _{(l d ', s d'} ), respectively (λ, ξ), (λ' , ξ') When is consistent with estimates of, and length It can be said that there is a high possibility that two line segments whose angles are (λ, ξ) and (λ′, ξ′) are connected. Therefore, the two pairs of line segments that constitute the contour of the object T at every length and angle estimates for a continuous matching _{(l d, s d),} (l d ', s d') a number If there is a large number (that is, if there is a preset number or more), it can be estimated that two line segments having the length and the angle are connected.

Furthermore, the sensor 20 _{(l d,} s _d), with respect to _{(l d ', s d'} ) When the sensing in the order _{of, (l d, s} d) _{is, (l d ', s d'} ) Thus, it can be seen that the object is ahead of the object T. _Thus, (l d, _{s d)} is a line segment having a length and angle to match _the, in the front of the line segment having a length and angle to match the _{(l d ', s d'} ), these It can be inferred that the line segment of is connected.

As described above, the connection relationship between two line segments among the line segments estimated by the first shape estimation unit 104 and the second shape estimation unit 105 (that is, whether or not they are connected, one line segment is It is estimated whether it is in front of the other line segment). Therefore, the entire contour of the object T can be estimated by estimating the connection relationship between the line segments.

Step S7: Next, when the target T has a recess, the correction unit 107 misses the estimation of the line segment that constitutes the contour of a part of the target T based on the sensing result r(t) of the recess. To correct. However, in step S7, when the target T is convex and the estimation is performed well, the correction by the correction unit 107 is not performed.

If the entire contour of the object T has not been estimated up to step S6, there is a possibility that the object T has a recess. When the concave portion is present, the line segment that is in front of the line segment that forms the concave portion may cause the sensor 20 to be unable to detect the line segment that forms the concave portion.

For example, the sensing result r ₄ (t) shown in FIG. 6 detects one line segment that constitutes the concave portion, but is before the target T at the time when r ₄ (t) changes discontinuously. The line segment makes it impossible to detect the line segment forming the recess. Thus, the sensor 20 _4, it is impossible to detect the entire line constituting the recess, corresponding to the line segment (l _{d, s d)} is not extracted.

As a result, the observed value of n _d (λ, ξ) is smaller than that in the case where the object T has no concave portion, but E[ _nd (λ, ξ) calculated by the above equations 4 and 5 is obtained. )] does not consider the concave portion (that is, assumes that the object T is convex), the number of line segments whose length and angle are (λ, ξ) is too small.

Therefore, when it is estimated that there are recesses on the object T, in order to compensate for this, instead of the equations 4 and 5 above, according to equation 10 below _{E [n d (λ, ξ} | ξ' )] as the expected value of n _d (λ, ξ). Here, ξ′ is the angle of the line segment that is adjacent to the line segment having the length and angle of (λ, ξ) and that constitutes the concave portion.

これにより、長さ及び角度がそれぞれλ及びξであると推定される線分の数が増加する（この増加には、０が１以上になる場合も含む）ことで、最終的に対象物Ｔの輪郭を構成する全ての線分を推定できることが期待できる。なお、ステップＳ７の補正を行った場合には、当該補正後に推定された線分も用いて、再度、上記のステップ６の接続関係の推定を行う。

As a result, the number of line segments whose length and angle are estimated to be λ and ξ, respectively, increases (this increase includes the case where 0 becomes 1 or more), and finally the target T It can be expected that all the line segments that make up the contour of can be estimated. When the correction in step S7 is performed, the line relationship estimated after the correction is also used to again estimate the connection relationship in step 6 above.

Here, f in the above equation 10 is given by the following. That is,

と定義するとき、０≦ξ_ｉ−１，ξ_ｉ＜２πに対するｆ（θ，ｘ）は、以下で与えられる。

, F(θ, x) for 0≦ξ _i−1 , ξ _i <2π is given by

ステップＳ８：出力部１０８は、最終的に得られた推定結果（すなわち、推定された線分及び推定された線分の接続関係）を出力する。なお、凹部の補正が行われた場合、推定された線分には、当該補正後に推定された線分が含まれる。

Step S8: The output unit 108 outputs the finally obtained estimation result (that is, the estimated line segment and the estimated line segment connection relationship). When the concave portion is corrected, the estimated line segment includes the line segment estimated after the correction.

As described above, the shape estimation apparatus 10 according to the embodiment of the present invention linearly moves in the monitoring area A using the sensing results r(t) of the plurality of sensors 20 randomly installed in the monitoring area A. The shape of the target T moving at a constant speed can be estimated. Moreover, the shape estimation device 10 according to the embodiment of the present invention estimates the shape of the target T using the sensing result r(t) of the sensor 20, which is a commercially available directional distance sensor. be able to.

<Effect of the present invention>
Here, the effect of the present invention will be described with reference to FIG. 7. FIG. 7 is a diagram illustrating an example of the effect of the present invention.

First, as a premise, the monitoring area A is assumed to be a rectangular area of 5000 in the x-axis direction and 300 in the y-axis direction.

FIG. 7A shows an object T and an actual contour of the object T. It is assumed that the object T moves along the straight line of y=150 at a speed of 1 in the x-axis direction. In addition, it is assumed that the number of sensors 20 randomly installed in the monitoring area A is 2000 and the maximum sensing range r _max of each sensor 20 is 100.

At this time, FIG. 7B shows a comparison between the shape estimated by the shape estimation device 10 according to the embodiment of the present invention and the actual shape of the object T. As shown in FIG. 7B, it is understood that the shape estimation device 10 in the embodiment of the present invention can estimate the shape of the target T with high accuracy.

However, since there are cases where two estimated values of the angle occur, and there is a case where it is not uniquely determined from the connection relationship, a plurality of shape estimations with small differences may be obtained. In addition, the start point and the end point of the estimated shape may not match. The estimated value of the speed was 0.95509.

The present invention is not limited to the above specifically disclosed embodiments, and various modifications and changes can be made without departing from the scope of the claims.

1 shape estimation system 10 shape estimation device 20 sensor 100 shape estimation program 101 sensing result management unit 102 speed estimation unit 103 extraction unit 104 first shape estimation unit 105 second shape estimation unit 106 connection relation estimation unit 107 correction unit 108 output Part 111 Temporary estimation part 112 Consistency judgment part 113 Adopting part 200 Sensing result storage part

Claims (8)

A plurality of directional distance sensors installed in a predetermined area, and an acquisition unit that acquires a sensing result of sensing an object that moves linearly at a predetermined speed in the area, respectively.
Extraction means for extracting the characteristic parameter of the first section included in the sensing result acquired by the acquisition means;
Estimating means for estimating a shape parameter of a second section included in the contour of the object and corresponding to the first section based on the characteristic parameter extracted by the extracting means;
Connection means for estimating the connection relationship between the shape parameters estimated by the estimation means,
A shape estimation device comprising: The extraction means is
Extracting a characteristic parameter indicating the length and direction of the first section, which is a line segment section included in the sensing result acquired by the acquisition unit,
The estimation means is
Based on the characteristic parameters extracted by the extraction means, a shape parameter indicating a length and a direction of a second segment, which is a line segment section included in the contour of the object and corresponds to the first section, is estimated. The shape estimation device according to claim 1, wherein The estimation means is
The provisional shape parameter indicating the length and direction of the line segment included in the contour of the object is estimated based on the first feature parameter and the second feature parameter among the feature parameters extracted by the extraction means. Then
The estimated provisional shape parameter determines the number of matches with other feature parameters,
The shape estimation device according to claim 1, wherein a predetermined number of the provisional shape parameters are set as the shape parameters in ascending order of the number based on the determined result. The estimation means is
Corresponding to the first sensing direction of the first directional distance sensor corresponding to the first sensing result in which the first characteristic parameter is extracted, and the second sensing result in which the second characteristic parameter is extracted. The provisional shape is obtained by solving a predetermined simultaneous equation in which the second sensing direction of the second directional distance sensor and the length and direction of the line segment section included in the contour of the object are unknowns. The shape estimation apparatus according to claim 3, wherein the parameter is estimated. The connection means is
The characteristic parameters of the two first sections, which are included in the sensing result and have at least one end connected, are specified, and the shape parameters that match the two specified characteristic parameters are connected at least one end. It estimates, The shape estimation apparatus as described in any one of Claim 1 thru|or 4 characterized by the above-mentioned. The correction means for correcting an error due to the concave portion of the number of shape parameters estimated by the estimating means when the object has a concave portion. Shape estimation device. An acquisition procedure in which a plurality of directional distance sensors installed in a predetermined area each obtain a sensing result of sensing an object that moves linearly at a predetermined speed in the area,
An extraction procedure for extracting the characteristic parameter of the first section included in the sensing result acquired by the acquisition procedure;
An estimation procedure for estimating a shape parameter of a second section included in the contour of the object and corresponding to the first section based on the characteristic parameter extracted by the extraction procedure;
A connection procedure for estimating the connection relationship between the shape parameters estimated by the estimation procedure,
A shape estimation method characterized in that a computer executes the method. A program for causing a computer to function as each unit in the shape estimation device according to claim 1.

Images