JP2857834B2 - Target position acquisition method and target position acquisition device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a target position obtaining method and a target position obtaining apparatus, and more particularly, to a target position obtaining method, even if a sub-target similar to the target exists near the target, without interfering with the sub-target. The present invention relates to a target position acquisition method and a target position acquisition device capable of acquiring a correct approximate position of a target.

[0002]

2. Description of the Related Art FIG. 6 shows an example of a conventional target position acquisition device. The target position acquisition device 600 captures a predetermined visual field and outputs a video signal E, and generates and outputs coordinates (H, V) and pixel values G of each pixel from the video signal E. And the image processing apparatus 20 which outputs the frame synchronization signal F, and the coordinates (Hs, Vs), (H
e, Ve) (see FIG. 7)
And a center-of-gravity acquisition device 630 that obtains the center of gravity (Hg, Vg) of the window region R from the coordinates (H, V) of the pixels within the pixel and the pixel value G. The center of gravity acquisition device 630
Has a center-of-gravity calculating circuit 62 for calculating the center-of-gravity position (Hg, Vg) according to Hg = {G · H} / {G} Vg = {G · V} / {G}.

FIG. 7 illustrates an image I in a certain frame. In this image I, a target T is photographed, and a sub-target J similar to the target T is photographed. The window region R is set to include the target T. The sub-goal J is outside the window region R. In addition,
It is assumed that the image of the target T is circular and its center coordinates are (H1,
V1). Also, assume that the image of subgoal J is also circular,
The center coordinates are (H2, V2). The coordinates of the diagonal points of the window area R are (Hs, Vs) and (Hs).
e, Ve).

FIG. 8 shows each axis H, V of coordinates (H, V) as 2
FIG. 3 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates with two horizontal axes and a pixel value G as a vertical axis. Assuming that the pixel value solid of the target T is symmetrical with respect to the center coordinates (H1, V1) and the pixel values of the parts other than the target T are “0”, the barycenter position (Hg, Vg) of the window region R = The center coordinates (H1, V1) of the pixel value solid of the target T are obtained. Therefore, the position of the center of gravity (Hg,
By calculating Vg), the approximate position of the target (T) can be obtained.

[0005]

As shown in FIG. 9, there is a case where the sub-target J approaches the target T and enters the window region R. FIG. 10 is a diagram corresponding to FIG. 8 in this case. If the pixel value solid of the sub-target J is not negligible compared to the pixel value solid of the target T, the position of the center of gravity (Hg, Hg, Vg) is calculated at an intermediate position between the two pixel value solids. That is, in the above-described conventional target position acquisition device 600, when a sub-target J similar to the target T exists near the target T, the sub-target J
There is a problem that it is not possible to obtain a correct approximate position of the target T. Therefore, an object of the present invention is to provide a target position obtaining method and a target position obtaining method capable of obtaining a correct approximate position of a target T without being misled by a sub-target J, even when a sub-target similar to the target exists near the target. It is to provide a position acquisition device.

[0006]

According to a first aspect of the present invention, a window region (R) including a target (T) is provided on an image (I), and coordinates of pixels in the window region (R) are provided. A target for obtaining the center of gravity (Hg, Vg) of the window area (R) based on (H, V) and the pixel value (G), and obtaining the center of gravity (Hg, Vg) as an approximate position of the target (T) In the position acquisition method, the gazing point coordinates (Hp, V
p) is set, the coordinates of the gazing point (Hp, V
In the vicinity of (p), the value becomes a relatively large value, and the gazing point coordinates (H
p, Vp), the pixel value (G) is multiplied by a weight (W) which becomes a relatively small value when the pixel value (G) is apart from the coordinates (H, V) of the pixel in the window area (R) and the weight / pixel value product. The center of gravity position (Hg, Vg) of the window area (R) is obtained from (WG), and the center of gravity position (Hg, Vg) is obtained.
Is obtained as the approximate position of the target (T).

[0007] In a second aspect, the present invention provides a method comprising:
Is provided on the image (I), and the coordinates (H, V) of the pixels in the window area (R) and the pixel value (G) indicate the position of the center of gravity (R) of the window area (R). Hg, Vg), and the position of the center of gravity (Hg, Vg)
In the target position acquiring device for acquiring the approximate coordinates of the target (T), a gazing point coordinate setting means (3) for setting gazing point coordinates (Hp, Vp), and the gazing point coordinates (Hp, Vp).
The distance (D) is small by the distance calculating means (4) for calculating the distance (D) from p) to the coordinates (H, V) of the pixel, and the distance (D) and a predetermined weighting function W (D). Weight generating means (5) for generating a weight (W) having a larger value; and calculating a weight / pixel value product (WG) by multiplying the pixel value (G) of the pixel by the weight (W). Multiplication means (1) for calculating the center of gravity position (Hg, Vg) of the window area (R) based on the coordinates (H, V) and the weight / pixel value product (WG). 2) A target position acquisition device characterized by comprising:

According to a third aspect of the present invention, in the target position acquiring apparatus having the above-mentioned configuration, when the image (I) is updated at predetermined time intervals, the gazing point coordinate setting means (3) includes:
The center of gravity position (H calculated for the image (I) at a certain time
g, Vg) are set as gazing point coordinates (Hp, Vp) for the image (I) at the next time.

[0009]

In the target position acquiring method according to the first aspect and the target position acquiring apparatus according to the second aspect, the value becomes relatively large near the set gazing point coordinates (Hp, Vp). Hp, Vp), the pixel value (G) is multiplied by a weight (W) which becomes a relatively small value to obtain a weight / pixel value product (W · G). The barycentric position (Hg, Vg) of the window region (R) is obtained from the coordinates (H, V) of the pixels in the window region (R) and the weight / pixel value product (WG), and the barycentric position (Hg) is obtained. , Vg)
As the approximate position of the target (T). Therefore, if the gazing point coordinates (Hp, Vp) are set in the vicinity of the target (T), the pixel value solid of the target (T) is lengthened, and the pixel value solid of the sub-target (J) is reduced. Then, the barycentric position (Hg, Vg) of the window region (R) is obtained. Therefore, the influence of the sub-target (J) is suppressed, and a correct approximate position of the target T can be obtained.

In the target position acquiring apparatus according to the third aspect, when the image (I) is updated at predetermined time intervals, the center of gravity (Hg, Vg) calculated for the image (I) at a certain time.
g) is changed to the gazing point coordinates (H) for the image (I) at the next time.
p, Vp). Thus, if the coordinates of the gazing point (Hp, Vp) are first set near the target (T), then, even if the target (T) moves, the coordinates of the gazing point (Hp, Vp) follow the target (T). Vp) moves, and the approximate position of the correct target T can always be obtained.

[0011]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in more detail with reference to the embodiments shown in the drawings. It should be noted that the present invention is not limited by this. FIG. 1 is a configuration diagram of an embodiment of a target position acquisition device according to the present invention. This target position acquisition device 1
00 denotes a camera 10 that captures a predetermined visual field and outputs a video signal E, and the coordinates (H,
V) and a pixel value G are generated and output, and the image processing apparatus 20 outputs a frame synchronization signal F, and a window area R defined by coordinates (Hs, Vs) and (He, Ve) of diagonal points. And a center-of-gravity acquisition device 30 for calculating the center of gravity (Hg, Vg) of the window region R based on the coordinates (H, V) of the pixels in (see FIG. 7) and the weight / pixel value product WG. I have.

The center-of-gravity acquisition device 30 includes a multiplication circuit 1
It includes a center-of-gravity calculation circuit 2, a CPU 3, a distance calculation circuit 4, and a weight generation circuit 5. The multiplying circuit 1 multiplies the pixel value G by the weight W to calculate a weight / pixel value product WG. The center-of-gravity calculating circuit 2 calculates the center-of-gravity position (Hg, Vg) according to Hg = {W · G · H} / {W · G} Vg = {W · G · V} / {W · G}. calculate.

The CPU 3 receives the coordinates of the gazing point (Hp, Vp) input by the operator, and inputs the coordinates of the gazing point (Hp).
(p, Vp) is passed to the distance calculation circuit 4 as an initial value.
When the position of the center of gravity (Hg, Vg) is calculated even once, the latest position of the center of gravity (Hg, Vg) is set to the gazing point coordinate (H
p, Vp) to the distance calculation circuit 4. Also, C
PU3 calculates the coordinates (Hs, V) of the diagonal point input by the operator.
s) and (He, Ve) are received and passed to the center-of-gravity calculation circuit 2 as initial values of the coordinates (Hs, Vs) and (He, Ve) of the diagonal points. And the position of the center of gravity (Hg, Vg)
, The movement amounts ΔHg and ΔVg are calculated.
ΔVg is set to the coordinates (Hs, Vs), (He,
Ve) (Hs + ΔHg, Vs + ΔVg),
(He + ΔHg, Ve + ΔVg) is passed to the center-of-gravity calculating circuit 2 as coordinates (Hs, Vs) and (He, Ve) of new diagonal points. Further, the CPU 3 determines a predetermined weight function W
Data for realizing (D) is written in the weight generation circuit 5. This will be described later in detail.

The distance calculation circuit 4 calculates D ・ D = (H-Hp) ・ (H-Hp) + (V-Vp) ・
From (V−Vp), the square D · D of the distance is calculated, and the distance from the fixation point to the pixel is obtained.

The weight generating circuit 5 calculates the square of the distance D · D
Is a look-up table in which a weight W is output and a weight W is output. A relatively large weight W is generated when the square of the distance D · D is small, and a comparison is performed when the square of the distance D · D is large. A weight W having an extremely small value is generated. As mentioned above, C
PU3 writes the data of the look-up table so as to realize the predetermined weight function W (D). Weight function W
As an example of (D), W = 255 · exp {−a · D · D}. In addition, W = 255 / (1 + a.D.D). FIG. 2 illustrates a graph of the weight function W (D).

FIG. 3 shows each axis H, V of coordinates (H, V) as 2
FIG. 3 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates with two horizontal axes and a pixel value G as a vertical axis. Sub-goal J
Is approaching the target T and is in the window region R. The image of the target T is circular, and its center coordinate is (H
1, V1). The image of the sub-target J is also circular, and its center coordinates are (H2, V2). FIG. 4 is a conceptual diagram in which the window region R is represented by orthogonal three-dimensional coordinates in which the axes H and V of the coordinates (H, V) are two horizontal axes and the weight W is the vertical axis. The weight W is the gazing point coordinates (Hp, Vp)
, And becomes smaller as the distance from the gazing point coordinates (Hp, Vp) increases. FIG. 5 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates in which each axis H, V of coordinates (H, V) is two horizontal axes, and the weight / pixel value product WG is a vertical axis. is there. Since the gazing point coordinates (Hp, Vp) are set near the center coordinates (H1, V1) of the target T, the pixel value solid of the target T is lengthened, and the pixel value solid of the sub-target J is dwarfed. ing. Therefore, the influence of the sub-target J is suppressed, and the position of the center of gravity (Hg, Vg) is near the center coordinates (H1, V1) of the target T. Thus, even when a sub-target J similar to the target T exists near the target T, a correct approximate position of the target T can be acquired without being misled by the sub-target J.

[0017]

According to the target position obtaining method and the target position obtaining apparatus of the present invention, even when a sub-target similar to the target exists near the target, the correct approximate position of the target can be determined without being misled by the sub-target. You can get it.
Therefore, it is particularly useful for a foreign substance inspection device, an intruder monitoring device, a visual sensor of a palletizing robot, and the like.

[Brief description of the drawings]

FIG. 1 is a configuration diagram of an embodiment of a target position acquisition device according to the present invention.

FIG. 2 is an exemplary diagram of a weight function.

FIG. 3 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates in which coordinate axes H and V are two horizontal axes and a pixel value G is a vertical axis.

FIG. 4 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates in which coordinate axes H and V are two horizontal axes and a weight W is a vertical axis.

FIG. 5 is a conceptual diagram showing a window region R by orthogonal three-dimensional coordinates in which coordinate axes H and V are two horizontal axes and a weight / pixel value product WG is a vertical axis.

FIG. 6 is a configuration diagram of an example of a conventional target position acquisition device.

FIG. 7 is an illustration of an image in a certain frame.

FIG. 8 is another conceptual diagram in which a window region R is represented by orthogonal three-dimensional coordinates in which coordinate axes H and V are two horizontal axes and a pixel value G is a vertical axis.

FIG. 9 is an illustration of an image in another frame.

FIG. 10 is still another conceptual diagram in which a window region R is represented by orthogonal three-dimensional coordinates in which coordinate axes H and V are two horizontal axes and a pixel value G is a vertical axis.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Multiplier circuit 2, 62 Centroid calculation circuit 3 CPU 4 Distance calculation circuit 5 Weight generation circuit 10 Camera 20 Image processing device 30, 630 Centroid acquisition device 100, 600 Target position acquisition device G Pixel value H coordinate (horizontal direction) V coordinate ( (Vertical direction) F Frame synchronization signal Hg barycentric coordinates (horizontal direction) Vg barycentric coordinates (vertical direction) W weight D distance Hp gazing point coordinates (horizontal direction) Vp gazing point coordinates (vertical direction) T target J sub-target R window area

Claims (3)

(57) [Claims] 1. A window region (R) containing a target (T).
Is provided on the image (I), and the barycentric position (Hg, Vg) of the window area (R) is obtained from the coordinates (H, V) of the pixels in the window area (R) and the pixel value (G), In a target position obtaining method for obtaining the center of gravity position (Hg, Vg) as an approximate position of the target (T), when the gazing point coordinates (Hp, Vp) are set, the gazing point coordinates (Hp, Vp) The pixel value (G) is multiplied by a weight (W), which becomes a relatively large value near and becomes a relatively small value away from the gazing point coordinates (Hp, Vp). The center of gravity position (Hg, Vg) of the window area (R) is obtained from the coordinates (H, V) and the weight / pixel value product (WG), and the center of gravity position (Hg, Vg) of the target (T) Target position acquisition characterized by being acquired as an approximate position How to get. 2. A window region (R) including a target (T).
Is provided on the image (I), and the barycentric position (Hg, Vg) of the window area (R) is obtained from the coordinates (H, V) of the pixels in the window area (R) and the pixel value (G), A target position acquisition device for acquiring the position of the center of gravity (Hg, Vg) as an approximate position of the target (T); a gazing point coordinate setting means (3) for setting gazing point coordinates (Hp, Vp); A distance calculating means (4) for calculating a distance (D) from (Hp, Vp) to the coordinates (H, V) of the pixel, the distance (D) and a predetermined weighting function W
(D) a weight generating means (5) for generating a weight (W) having a larger value as the distance (D) is smaller, and a weight obtained by multiplying the pixel value (G) of the pixel by the weight (W). A multiplication means (1) for calculating a pixel value product (WG), and a barycentric position (Hg) of the window region (R) based on the coordinates (H, V) and the weight / pixel value product (WG) , Vg)
And a gravity center position calculating means (2) for calculating the target position. 3. The target position acquisition device according to claim 2, wherein when the image (I) is updated every predetermined time, the gazing point coordinate setting means (3) sets the image (I) at a certain time.
A target position acquisition device characterized in that the center of gravity position (Hg, Vg) calculated for is set as the gazing point coordinates (Hp, Vp) for the image (I) at the next time.