JPS61205807A - Actual length conversion system of visual sensation sensor - Google Patents

Abstract

PURPOSE:To make it possible to perform the conversion of an actual length by automat ically determining an optimum binarization threshold valve, by determing a binarization threshold value so that the difference between the number of picture elements to two figures having an equal area becomes min. CONSTITUTION:Two figures of threshold value determining patterns are binarization in picture element number determination parts 101, 102 by using the threshold value (t) generated in a threshold value generation part 103 to respectively generate the numbers S1, S2 of picture elements. A first operation part 104 calculates the difference between the numbers S1, S2 of picture elements and controls the threshold value genera tion part 103 so as to bring the same to a min. value to generate an optimum threshold value from the generation part 103 and, at the same time, to judge that the difference between the numbers S1, S2 of picture elements is a min. value in the change range of the threshold value (t). A second operation part 105 calculates conversion coefficient from the numbers S1, S2 of picture elements at this time and the actual area of the patterns. The actual length of an article to be measured can be calculated by multiply ing the size of the image on a visual sensation sensor by this conversion coefficient.

Description

[Detailed Description of the Invention] [Summary] When the image on the visual sensor is binarized and multiplied by a conversion coefficient to obtain the actual length, a pattern consisting of two figures having the same area is used, and the image on the visual sensor is By setting the threshold for binarization so that the difference in the number of pixels between both figures is the minimum value, the optimal threshold is determined regardless of how the screen looks, and the actual area of the pattern and this threshold are used. A conversion coefficient is determined from the sum of the areas of both figures obtained by the method, and the conversion coefficient is used to convert the image on the visual sensor and the actual length.

[Industrial application field]

The present invention relates to a method for converting information from a visual sensor into actual length, and in particular, when controlling a robot using information from a visual sensor, converting the size of an image captured by a visual sensor into actual length. The present invention relates to a method for converting the actual length of a visual sensor to teach the robot.

When using a visual sensor as the robot's eyes and controlling the robot using the information obtained from it, the image on the screen of the visual sensor is binarized to determine the size of the image and converted to this value. Conversion to actual length is performed by multiplying by a coefficient, but if the binarization threshold at this time is inappropriate, the size of the recognized image will change, and therefore conversion to actual length may not be performed correctly. Can not. Therefore, the optimal 2
There is a need for a method that can automatically determine the valuation threshold and perform actual length conversion.

(Conventional technology) Conventionally, the threshold value in image binarization processing when converting information from a visual sensor into actual length was determined manually by cut-and-try methods depending on how the screen looked. .

[Problem that the invention seeks to solve]

However, this method is undesirable because not only are there individual differences, but it also takes an extremely long time to finally determine the optimal value. There was a need for a method that would allow conversion factors to be determined.

[Means for Solving the Problems] FIG. 1 shows the configuration of the system of the present invention, and the two figures of the threshold value determination pattern are the threshold values generated by the threshold value generation unit 103 in the pixel number determination unit 101 and 102. The pixels are binarized using t to generate the number of pixels s, , s2, respectively. The first calculation unit 104 calculates the difference between the numbers of pixels s, s2, and controls the threshold generation unit 103 so that the difference becomes the minimum value. It is determined that the difference between the number of pixels s, , s2 is the minimum value in the range of change of . The second calculation unit 105 calculates a conversion coefficient from the number of pixels S, s2 at this time and the actual area of the pattern. The actual length of the object to be measured is determined by multiplying the size of the image on the visual sensor by this conversion coefficient.

[For production]

In the method of the present invention, a binarization threshold is set so that the difference in the number of pixels between two figures having the same area is minimized, and the area of the image binarized by this threshold is calculated from the actual area of the pattern. Since the actual length of the object to be measured is determined by determining the conversion coefficient, the actual length can always be converted correctly regardless of how the image appears on the screen of the visual sensor.

〔Example〕

FIG. 2 shows a threshold value determination template 1 in an embodiment of the method of the present invention, which has a ring shape with an inner radius rl and an outer radius r2, and an outer ring portion 2 shown with diagonal lines. The radii rI and r2 are selected so that the area with the inner opening 3 within the ring is equal. That is, the radius rI
, r2 shall satisfy the following relationship.

2πr, 25πr22 FIG. 3 shows the configuration of an embodiment of the present invention, in which 11 is a visual sensor camera, 12 is a monitor TV (television), 13 is an image processing processor, 14 is a frame memory, 15 is memory, 16 is microcomputer, 17
18 is a keyboard, 19 is an external storage device, and 20 is an interface circuit.

In the method of the present invention, a template 1 as shown in FIG. 2 is placed in front of a camera 11 in a visual sensor, and the inner opening 3 is made visible as a hole within the field of view of the visual sensor via a monitor TV 12. .

In this state, the image processor 13 binarizes the image using the threshold given by the microcomputer 16 and stores it in the frame memo January 4. At this time, keyboard 1
The initial value of the threshold can be given via 8. The microcomputer 16 performs calculations using the image data accumulated in the frame memo January 4, and calculates the number of pixels of the outer ring part of the template 1, 82 (1), and the number of pixels of the inner part, 5t (t).
Find and output. The calculation unit 13 has 32 pixels (t)
The difference D (t) between and the number of pixels 81 (t) is calculated. That is, the difference D(t) is D(t)=ls2(t)-s,(t)1.

At this time, the microcomputer 16 changes the threshold value of the binarization process in the image processing processor 13, and
If the difference D(to) at o satisfies p(to)<ε with respect to the constant ε, and D(to) becomes the minimum value of the difference D(t) in the vicinity of t=t o, that is, D ( to
) -Minimum D (t)fD-L t-' A threshold value tO such that the tongue is 49 is determined and output. The threshold tO is the optimal threshold as described above.

The microcomputer 16 uses the number of pixels 82 in the outer ring portion and the number Sl of pixels inside the circle corresponding to the optimal threshold value to obtained in this way to calculate the conversion coefficient between the image and the actual length according to the following equation.

Conversion coefficient A-E7τs, +s-go (m/ pixel
) Here, S is the actual area of the outer ring and the inside of the circle (
It is Tsuru 2).

If the conversion coefficient A is determined in this way, the actual length of the actual object to be measured can be determined as L=A-Lg. Here, Lg is the length of the visual sensor on the screen. The obtained actual length is given to a robot or the like (not shown) via the interface circuit 20. This value can also be displayed on the screen of the CRT 17, for example.

FIG. 4 shows the procedure of conversion coefficient calculation processing in the method of the present invention. That is, first, the initial threshold T min
is set from the keyboard, etc., and the areas S r (T) and S 2 (T) of the template at the threshold T are measured to obtain the error D (T)''132 (t) −3+ (t)1.Next, The same calculation is repeated by adding 1 to the threshold value T to determine the threshold value TO that minimizes the difference D (T) by going up to the maximum threshold value T IWax.Next, the error D (TO) due to this threshold value To is calculated from the constant ε. If it is smaller than ε, change the threshold Tm1n and Tmax+ε and repeat the calculation, and if it is smaller than ε, calculate D( T), and after determining that the error D(TI) at that time is the minimum value at the threshold T, if the error D(TI) is smaller than the constant ε, set the optimal threshold to T. , and thereby determine the conversion coefficient.If the error D(TI) is not smaller than the constant ε, it is treated as an error.

In this way, according to the method of the present invention, by determining the threshold so that the difference between the number of pixels between the outer ring and the inside of the circle is the minimum value, binarization is always performed under constant conditions regardless of how the image looks. A threshold value can be determined, and therefore, the image of the visual sensor and the actual length can always be correctly converted using the conversion coefficient corresponding to the threshold value thus determined.

The template for threshold value determination in the method of the present invention is not limited to the shape shown in FIG. 1; for example, in FIG.
As shown in (b), it may be a different shape such as a rectangle or a triangle with a window inside, or it may be a line of shapes having the same area as shown in +8) of the same figure. In this case, it is desirable that the two figures to be compared when determining the threshold value be colored with strong contrast, such as if one is black, the other is white.

〔Effect of the invention〕

As explained above, according to the method of the present invention, a binarization threshold is determined so that the difference in the number of pixels for two figures having the same area is minimized, and the area of the image binarized by this threshold is Since the conversion coefficient is determined from the actual area of the pattern and the actual length of the object to be measured, the actual length can always be converted correctly regardless of how the image looks on the screen of the visual sensor.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of the method of the present invention; FIG. 2 is a diagram showing a template for threshold determination in an embodiment of the method of the present invention; FIG. 3 is a diagram showing the configuration of an embodiment of the method of the present invention; FIG. 4 is a diagram showing the procedure of the conversion coefficient calculation process in the method of the present invention, and FIG. 5 is a diagram showing another example of a template for determining a threshold value. DESCRIPTION OF SYMBOLS 1--Template, 2--Outer ring part, 3--Inner part, 11--Camera, 12--Monitor TV (television), 13--Image processing processor, 14--Frame memory, 15--Mesori , 16--Microcomputer, 17--CRT (cathode ray tube), 18--Keyboard, 19--External storage, 20-Interface circuit Diagrams showing the configuration of the system of the present invention.

Claims (1)

[Claims] In a method of binarizing an image on a visual sensor and multiplying it by a conversion coefficient to obtain the actual length, the difference in the number of pixels in the image on the visual sensor for a pattern consisting of two figures having the same area. means for determining an optimal threshold for binarization so that the value becomes a minimum value; and means for determining the conversion coefficient from the actual area of the pattern and the sum of the areas of both figures determined by the optimal threshold. An actual length conversion method for a visual sensor, characterized in that the actual length of an image on the visual sensor is determined using the determined conversion coefficient.