JPS60201778A - Picture input device - Google Patents

Abstract

PURPOSE:To control the posture of a camera so that the center of camera viewfield is coincident with the position of an object, by picking up the object together with measurement of the input light quantity of an image pick-up means and controlling the posture of the image pick-up means according to the result of measurement of said light quantity. CONSTITUTION:A camera 1 is set at an upper oblique position to an object 4 on a table 5, and a light source is set so that its optical axis is coincident with the visual axis of the camera 1. A post 61 is supported rotatably on a base 6 and at the same time an attachment shaft 62 of the camera 1 is attached tiltably at the upper end of the post 61. The picture signals given from the camera 1 are supplied to a light quantity measuring device 4 for measurement of the input light quantity to the camera 1. Based on the result of this measurement, a posture controller 8 sends a control signal (e) to the camera 1. At the same time, a posture information signal (f) is fed back from the camera 1. Then the turning amount theta of the post 61 and the tilting amount phi of the shaft 62 are prescribed for control of posture of the camera 1.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention The present invention relates to an image input device mainly used to image a three-dimensional object, for example in an object recognition device for an industrial robot.

<Background of the Invention> As shown in FIG. 9, a conventional image input device of this type has a fixed camera 9 whose viewing range A1 covers the entire table 5, and a limited viewing range A2 necessary for recognizing an object 4. The fixed camera 9 detects the position of the object 4 on the table 5, and based on the detection result, the posture control device 91 is activated to move the movable camera 90. The center of the field of view is set at the position of object 4. Detection of the position of the object by the fixed camera 9 is performed by calculating the density distribution Lx, Lγ of the image in the +Y direction on the camera imaging surface 92 (shown in FIG. 10) using the following equation.

L x = Jf (x, y)dxLy = ff
(x, y) dy where f(x, y) is the density of the image at the coordinates (x, y).

However, in the case of the conventional device described above, two cameras9.9
Since 0 is required, the equipment cost is high and the installation space for the equipment is large.

Furthermore, when the object image G obtained by the fixed camera 9 has a low contrast, the density distributions Lx and Ly are not particularly sharp, which may make it difficult to detect the position of the object.

Furthermore, when detecting the object position, the filtering degree distribution L in both x and 7 directions is
Since it is necessary to calculate x and Lγ, the processing becomes complicated,
There were a number of disadvantages, such as an increase in the size of the control device and higher costs.

<Object of the invention> The present invention can detect objects without using a camera for detecting object positions.
It is an object of the present invention to provide a novel image input device that can control the posture of an object recognition camera so that the center of its field of view is located at the object position, and thereby aims to eliminate the above-mentioned disadvantages at once.

<Structure and Effects of the Invention> In order to achieve the above object, the present invention provides a retroreflector behind an object, and implements coaxial illumination that irradiates the object with light along the visual axis of the imaging means. In this way, the object is imaged, the amount of light input to the imaging means is measured, and the attitude of the imaging means is controlled based on the measurement results.

According to the present invention, since a camera for detecting the position of an object is not required, equipment costs can be reduced and the installation space for the equipment can be reduced. Furthermore, by using a retroreflector and implementing coaxial illumination, a high-contrast image can be obtained, so the object position can be detected reliably. Moreover, since the object position is detected based on the input light intensity of the imaging means, compared to the conventional method of calculating the concentration distribution, it is possible to simplify the processing, downsize the control device, and reduce costs, thereby achieving the purpose of the invention. Achieve remarkable effects.

<Explanation of Examples> FIGS. 1 and 2 show the basic configuration of an image input device,
A CCD (Charged-Cou) that obtains an image of the object 4.
A camera 1 in which a device (Pled Device) was used;
It consists of an illumination light source 2 that irradiates light (indicated by an arrow in the figure) to an object 4, and a retroreflector 3 placed on the back of the object 4. This retroreflector 3 is an object that reflects light in the direction of incidence of the light, and is a sheet-like object affixed to a table 5, on which the object 4 is positioned.

The camera 1 is placed obliquely above the object 4 on the table 5, and the light source 2 is positioned so that its optical axis 20 coincides with the viewing axis 10 of the camera 1. In this embodiment, in order to realize this coaxial illumination, the 71-fmirror 21 is tilted and arranged on the visual axis 10 of the camera 1, and the light source 2
is positioned 16 toward the half mirror 21 and facing the visual axis 10.

According to the above method, the image characteristics of the camera 1 are a high contrast bimodal histogram with a large density difference between the object and the background, as shown by the solid line in Figure 3, and therefore, when binarizing image data, a threshold The value setting range H becomes larger. In the figure, the broken line is a histogram of low contrast uniformity for which it is difficult to set the threshold value.

FIG. 4 shows a specific configuration example of an image input device according to the present invention. In the illustrated example, a support 61 is rotatably supported on a base 6, and a mounting shaft 62 of the camera 1 is tiltably attached to the upper end of the support 61. In addition, in the figure, the light amount measuring device 7 inputs the image signal lij from the camera 1, and
The input light amount χi of camera' 1 is measured. The attitude control device 8 sends a control signal C based on the measurement result to the camera 1, and also feeds back an attitude information signal f from the camera 1 to define the rotation amount θ of the support column 61 and the tilt amount ψ of the mounting shaft 62. to control the attitude of the camera 1.

FIG. 5 shows an example of the configuration of the optical tj measuring device 7. As shown in FIG.

In the illustrated example, the image signal lij from the camera 1 is binarized for each pixel using a predetermined threshold value C, and this is added over all pixels to determine the input light amount λi of the camera 1. A comparator 71 in the figure compares the image signal 1ij with a threshold value C set within the range H shown in FIG. and outputs a binary signal B of logic "0". The adder 72 receives the binary signal B
is added over all pixels, and the input light amount λi is calculated by latching the addition output ΣB in the latch circuit 73 and adding it to the latch output ΣBp fed back with the binarized signal B. This input light amount χi is
The maximum input light intensity λ0 is maximum when there is no object within the field of view, and if the imaging surface of the camera 1 is composed of n-bit pixels in the vertical and horizontal directions, all pixels will be logic "1".
It is expressed as follows.

χQ = n
The input light amount χi is expressed by the following equation.

λi = n x n −S Therefore, if the input light amount χi is minimized, the object can be included in the field of view of the camera 1 to the maximum extent, and the attitude control device 8 controls the camera so that the input light amount χi is minimized. Control the posture of 1.

FIG. 6 shows another embodiment. In the illustrated example, an arm 60 is projected above a table 5, a base 6 is fixed to the lower surface of the arm 60, a support frame 63 is swingably attached to the base 6, and a camera 1 is attached to the support frame 63. It is rotatably mounted.

FIG. 7 illustrates a method for determining the field of view of the camera 1 with respect to the table 5. Now, if the field of view of camera 1 is set to 1/4 of the table area, first divide this table into 4 equal parts, and find the center point (points indicated by n = 1 to 4) and the points between each center point. An intermediate point (points indicated by n=5 to 9) is assumed. Next, set the viewpoint of camera 1 at each point n=l~4 in sequence,
The respective input light amounts χin (where n=l to 4) are measured. Then, a point that satisfies the following equation is detected, and based on the detection result, the visual field determination table shown in FIG. 8 is referred to to determine the viewpoint on table 5.

λin(λ.−ε However, ε is an error tolerance value.

For example, if it is assumed that the point that satisfies the above equation is n=1.2, then referring to the table in FIG. 8, the viewpoint of camera 1 will be determined to be n=5.

However, when coaxial illumination is applied to the object 4 from the direction of the camera 1, the shadow 41 that appears behind the object 4 is hidden by the object 4 and does not enter the field of view of the camera 1, and is therefore not subject to image input. Further, the light irradiated onto the object 4 is diffusely reflected on the object surface, and a part of the reflected light is incident on the camera 1. On the other hand, since the light irradiated to the retroreflector 3 is reflected toward the camera 1, most of the light is incident on the camera 1. As a result, the object 4 is imaged darkly and the background is brightly imaged, and a high-contrast image is obtained. From this high-contrast image,
The object position can be detected reliably and the field of view of the camera 1 can be set optimally.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the basic configuration of the image input device of the present invention, FIG. 2 is a diagram showing the positional relationship between the light source and the camera, FIG. 3 is a characteristic diagram showing the density distribution of the image, and FIG. FIG. 5 is a circuit block diagram of a light amount measuring device; FIG. 6 is a diagram showing a specific configuration of another embodiment;
FIG. 7 is a diagram showing the camera field of view determination method, FIG. 8 is a diagram showing the field of view determination table, FIG. 9 is a diagram showing the configuration of the conventional method, and FIG. 10 is a diagram showing the object position detection method in the conventional method. be. 1... Camera 2... River/Light source 3...
Retroreflector 7... Light amount measuring device 8...
...Posture control device patent applicant Tateishi Electric Co., Ltd. 3 ■ A/ 4 娃6 口

Claims (1)

[Claims] An imaging means for obtaining an image of the object, an illumination means for irradiating light onto the object along the visual axis of the imaging means, a retroreflector to be placed behind the object, and measuring the amount of light input to the imaging means. An image input device comprising: a measuring means for controlling the image capturing means; and an attitude controlling means for controlling the attitude of the imaging means based on the measurement results.