JPS61241611A - Visual sensor - Google Patents

Abstract

PURPOSE:To distinguish a rectangle surely by simple constitution by finding the length from the intersection of two diagonal lines to each apex and recognizing as a rectangle when the length is within a specified range. CONSTITUTION:A visual sensor controlling device 1, a higher rank controller 2, a monitor TV 5 a light pen etc. are provided. For operation mode of the device 1, a teaching mode that registers internal setting of the device 1 and objective work to be recognized and an on-line mode that performs measuring according to command of the controller 2 and reports the result of measurement to the controller 2. These modes are designated from picture of the TV 5 by the pen 6 and selected. In the teaching mode, picture of objective work 3 is read by a camera 4 and outputted to a picture inputting circuit 9. In the circuit 9, video signals are binary coded, and isolated point is removed and outputted to a picture storing circuit 11. In the on-line mode, a CPU 7 gives command to the circuit 9, reads the picture of the work 3 and prepares a stroke table.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a visual sensor that identifies and measures workpieces in, for example, assembly work or transportation work in factory automation.

[Conventional technology]

In recent years, robots and various types of equipment with advanced functions are being introduced to automate production processes in factories.
Since the shape and position of the target workpiece are not fixed, it is not possible to respond to changes in the shape and position, and there are still many fields that have to rely on human labor, such as assembly processes, but in order to fully automate these processes, Visual sensors are essential.

Conventionally, visual means (visual sensors) in such cases have been able to detect characteristics such as area, circumference length, hole area, maximum length and minimum length from the center of gravity of multiple workpieces that exist separately on the monitor screen. Equipped with a function that uses data to identify which workpiece corresponds to the taught (teaching operation), or to identify designated shapes such as circles, ellipses, and rectangles from the window area on the screen. is used.

[Problem that the invention seeks to solve]

In the conventional visual sensor described above, when identifying a rectangle, not only the length of the four line elements that make up the entire rectangle, but also the
The angle between each line element was also checked, which required complicated processing.

The present invention has been made with attention to this point, and an object thereof is to provide a visual sensor system that can reliably identify rectangles among the figures of a target workpiece with a simple configuration.

[Means for solving problems]

The visual sensor according to the present invention includes a camera that reads an image of a target work, and a video signal ll-2 from this camera.
An image input circuit that converts into values and removes isolated points; an image processing circuit that generates run lengths from output signals from this image input circuit; and stores the binarized skin binary image and recognition results. The image storage circuit, the monitor TV that displays the above recognition results, and the target work that should be recognized at least in the teaching mode? A light pen to register,
A storage circuit that stores the teaching data obtained during the teaching mode, and a central processing unit that is connected to the upper controller via an interface, exchanges measurement commands and measurement results with the upper controller, and controls all of the above circuits. It is equipped with the following.

[For production]

In this invention, the rectangle as the figure of the target work is the starting point of 54 line elements, fc, is the ending point, all m points, and the length from the intersection of the diagonals of the two trees to each vertex is calculated, and this is within a certain range. It is designed to be identified by the fact that

[Embodiment of the Invention] Figures 1 to 5 all show an embodiment of the present invention, and Figure 1 is a block diagram of the configuration of the visual sensor of the present invention;
2 to 5 are operation explanatory diagrams. First, in Fig. 1, (1) is a visual sensor control device, (2) is a host controller that gives measurement commands to this visual sensor control device 1 (11), (3) is a workpiece to be measured, and (4) is a The image of the target work (3) is a camera that reads and converts it into a video signal, and (5) is a monitor TV that displays the original image, binarized image, recognition results, and guidance for various operations. .

(6) is a light pen, and is it the internal settings of LeW and the target work to be recognized? This is used as a man-machine interface in the teaching mode for registration. (7)
reads the command re in the visual sensor control device (1), and this? A central processing unit (8) is a storage circuit that stores programs and teaching data output from the central processing unit (7), and (9) is a binary image from a video signal tube input from a camera (4). 0@ is an image processing circuit that generates run length data based on the signal output from this image input circuit (9), and 0D is an image Binary image from input circuit (9)? 'nz is an interface circuit with the monitor TV (5), an interface circuit with the 031 halide pen (6), and (14) is the image storage circuit that stores recognition results. This is a higher-level interface circuit that sends and receives commands and measurement results.

The visual sensor of this invention is configured as described above,
The operation will be explained based on FIGS. 2 to 5.

The operation modes of the visual sensor control device (1) include a “teaching mode” in which the internal settings of the visual sensor control device 1 (lj) and the target work to be recognized are registered, and a “teaching mode” in which measurement is performed by commands from the host controller (2). It has an "online mode" in which measurements are carried out and the measurement results are reported to the host controller (2), and each of these modes can be
It is possible to select from the screen (5) by giving an instruction using the light pen (6). Next, teaching mode and online mode will be explained separately.

First, in the "teaching mode", the image s'f of the target workpiece (3) is converted into a video signal to be read by the camera (4) at regular intervals and sent to the image input circuit (9).
Output to. Then, in this image input circuit (9), the video signal? In addition to binarizing, isolated points are removed, and the image storage circuit (1
Output to 0. The original image, binary image, etc. can be displayed on the monitor Tl5) by selecting one image, but the operator converts the workpiece N02 into binary data corresponding to the target workpiece (3) while watching the screen of this monitor TV(5) Types of local features such as level, circle, ellipse, rectangle, etc., and window setup from the monitor TV (5) screen to the light pen (6)
This is done by using this monitor TV (
5) Is the 7'c binary image completely displayed on the screen? While watching, you can also point out points of interest regarding local features.

Next, FIG. 2(a) shows an example of the above-mentioned binarized image. After specifying the rectangle of interest (161?s1 point (+)), use the light pen (6) to instruct the start of tt measurement.
) to give commands. This image input circuit (9) binarizes the video signal of the target work (3) read by the camera (4) "f!-" at a set binarization level, and also removes isolated points. , to the image processing circuit 00) to generate run-length data.Furthermore, this central processing unit (7) determines from the run-length data an area where run lengths are continuously connected, and calculates the area of this determined area. Create a line element table by racking outside [r].

For example, the left side of the area can be determined as follows.

In other words, the left end point of the run length of the beginning of the corresponding area or the cutting point (the run length is the starting point) is set as 1i″f[x(o), y・(0)1], and this position is When the position of the left end point of the run length of the pipe opening from n is set to t Cx (n), y (phantom), the straight line passing through those two points, that is, corresponds to y (1) to y (n-1). Buy X place xc (1
)-xo(n-1), the end point of the run length [x(1).

The length in the X direction between y(1)) and (x (n-1), y(n-1)) is l X o(i)-X(ll l <ΔX, where 1=1
~n-1 ΔX; If the allowable value is satisfied, repeat the same process for the next run length. And lXc(i)-X(i)l)ΔX1=1~n-1
Tonarare is (x(n)*y(n) with 1 as the cutting point,
Cx(o) l y(o3:] to (x(n-1), y
(n-1)]? The starting point and end point of one line element are registered in the line element table along with the slope.

As shown in FIG. 4, this line element table is configured to include data on the starting point position, ending point position, and slope for one line element.

Next, the point of interest (+) of the rectangle specified on the screen of the monitor TV (5) is determined from the line element table of the area inside the meeting,
In order to be rectangular, the corresponding area must be composed of four # elements, and in FIG. 2(b), the line elements Ll to Li+5 correspond. Then, the coordinates of the starting point or ending point of each line element are the vertices of the entire rectangle, and (xi, yi)
(xi+1.yi+1) (xi+2.yi+2) and (
xi+3. yt+5).

Next, find the intersection point of the two object lines connecting the four vertices mentioned above, and if this is f (xdc, ydc), the length tt - tA + 5 from the intersection point to each vertex is /j - (xj
-xdc)”+(yj-ydc)2(j=i ~i+5
) and 1-, so if they satisfy the following conditions, they can be identified as rectangles.

1 7 floats 1 ku α However, j = 1 ~ 4 α; Tolerance Therefore, as teaching data that defines a rectangle, take the average value of the two opposing sides, and calculate this by the long side (LL) and short side (Ls ) and register it in the storage circuit (8) together with the binarization level and window range.

Next, in the "online mode", the visual sensor control device (1) receives a measurement command (
As shown in Figure 5 (a), there is a work number designation)
When the input is made through the entire host interface circuit (141?), the central processing unit (7) receives this measurement command, sets the window range and binarization level from the teaching data, and sends the data to the image input circuit (9). Give the command, capture the entire image of the target workpiece (3) from the camera (4), run length with the image processing circuit (QO)? Generate 7'C# and line element table in a similar way to teaching mode? create. In this online mode, since the point of interest is not specified on the screen of the monitor TV (5),
Based on the teaching data from the first line element, a rectangle is identified in the following steps.

(11 When the line length of the focused line element is ?L, 1.1 <
: β or l 111 β However, β; Allowable value? Find a satisfying line element tk.

(A region containing line elements tk f?! The number of line elements forming the area is 4.

(2) Length of line element tk or long side L of teaching data
If the length of line element tk+1 corresponds to the short side L (on the contrary, the long side La), the length of the line element tk+1 is the above (1'
) all conditions must be satisfied.

■ Obtain the lengths from the intersection of diagonals to each vertex using the same method as during teaching for the line elements of the 4-tree tree, and ensure that the lengths are within the allowable value for the average value.

If there is a line element that satisfies the procedure (I) ~ @ described above, it is recognized as a rectangle, and the lengths of the long side and short side are calculated and LL'
, L8', and at the same time, the angles of the center and long sides of the rectangle with respect to the X axis are set as the position and direction of the local feature, respectively, to the upper controller (2) as shown in Fig. 5(b).
The report shall be as shown below.

In the above-described embodiment, a case has been described in which a whole rectangle is identified as an area, but a rectangle as a hole within an area can also be easily identified by switching between a positive image and a negative image of the image.

〔Effect of the invention〕

As described above, according to the present invention, the lengths from the intersection of two diagonals to each vertex are determined by taking all vertices as the starting point or ending point of four line elements, and making sure that these lengths are within a certain range. Since it is recognized that if there is a rectangle, it is recognized as a rectangle, it has an excellent effect of being able to reliably identify all rectangles with an extremely simple configuration.

[Brief explanation of the drawing]

Fig. 1 is a block diagram of the visual sensor of the present invention;
Figures (a) to (C) are binary images of the target workpiece, explanatory diagrams for recognizing rectangles from the lengths from the intersections of line elements and diagonals to each vertex, and Figure 6 shows the run of the area of the target workpiece. An explanatory diagram, Figure 4 is an explanatory diagram of the line element table, and Figure 5 (&
) and (b) are explanatory diagrams showing measurement commands and measurement results from a host controller. In the figure, (1) is a visual sensor control device, (2) is a host controller, (3) is a target workpiece, (4) is a camera, (5) is a TV monitor, (6) is a light pen, and (7) is a central processing unit, (8) a memory circuit, (9) an image input circuit, 00) an image processing circuit, (11) an image storage circuit,
(12 is the monitor interface circuit, am is the light pen interface circuit, and I is the upper interface circuit. In the figure, the same reference numerals indicate the same or corresponding parts. Agent: Patent Attorney Mitsuro Kimura Figure 2 (b) ) (C) "l" LV1+21 Figure 4 Figure 5

Claims (1)

[Claims] A camera that reads an image of the target workpiece, an image input circuit that binarizes the video signal from this camera and removes standing points, and an image processing circuit that generates a run length from the output signal from this image input circuit. an image storage circuit for storing the binarized binary image and the recognition results; a monitor TV for displaying the recognition results; a light pen for registering a target work to be recognized at least in teaching mode; A memory circuit that stores teaching data obtained during the mode, and a central processing unit that is connected to a higher-level controller via an interface and that sends and receives measurement commands and measurement results to and from the higher-level controller and controls each of the above circuits. A visual sensor characterized in that the central processing unit is configured to identify a rectangle as a figure of the target work.