JP2841961B2 - Visual recognition 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 visual recognition device which is installed in a manufacturing or inspection line, for example, in a factory, and is used for inspecting the appearance and shape of a product.

[0002]

2. Description of the Related Art In general, a visual recognition device takes an image signal obtained by capturing an image of an object by a television camera, performs preprocessing such as binarization on the image signal, and then converts the area of the binary image. , The center of gravity, the principal axis angle, etc.
By comparing the measured value with a preset reference value, the appearance and shape of the object are determined. In the image processing apparatus, a window for designating a measurement region is generally set for a target portion of a binary image, and a feature amount is measured for an image portion in the window.

[0003]

However, since the object is transported to an observation position by a belt conveyor or the like,
During transport, the position or orientation of the target object changes, and the target object may be out of the observation field of view of the television camera, or the target portion of the image of the target object may protrude from the window. Therefore, the observation field of view of the TV camera is set sufficiently wide, and the moving position of the image of the target object is calculated, and the window setting position is changed by the amount of movement, but the observation field of view of the TV camera is set wide. In addition, the measurement accuracy decreases, and high-precision inspection becomes impossible.

[0004] The present invention has been made in view of the above problem, and even if the position or orientation of an object changes, a visual recognition device capable of performing high-accuracy inspection without lowering measurement accuracy. The purpose is to provide.

[0005]

In order to achieve the above object, a visual recognition device according to the present invention comprises a first imaging means capable of including the entirety of an object within an observation field of view, and a portion of interest of the object. Only the first that can be included in the observation field of view
A second imaging unit having a higher resolution than the first imaging unit, a displacement calculating unit for calculating a displacement amount of the image of the object obtained by the first imaging unit, and an image obtained by the first imaging unit Correction value calculation means for calculating a correction value of a parameter for processing an image obtained by the second imaging means from the image processing apparatus, coordinates of an image obtained by the first imaging means, and an image obtained by the second imaging means The position shift amount calculated by the position shift calculating means using the dimensional ratio with the coordinates of
A displacement conversion means for converting the displacement of the image of the object obtained by the imaging means to an amount of displacement, and an image for processing the image obtained by the second imaging means by the correction value calculated by the correction value calculation means. Processing means, and measurement processing means for performing a predetermined measurement process by changing a set position of a measurement region by an amount of displacement converted by the displacement conversion means for an image obtained by the second imaging means. I have it.

[0006]

According to the present invention, the amount of displacement of an object is measured from an image from a first image pickup means having a wide observation field of view, and the amount of displacement of an image from a second image pickup means having a high resolution is converted from the measured value. Therefore, even if the position or orientation of the target object changes, high-precision inspection can be performed without lowering the measurement accuracy. In addition, since the correction value of the parameter for image processing is calculated based on the image obtained by the first imaging unit and the image obtained by the second imaging unit is processed by the correction value, even if there is a variation in illumination or the like, In addition, stable measurement is possible, and the frequency of inspection errors is reduced.

[0007]

FIG. 1 shows the appearance of a visual recognition device according to an embodiment of the present invention.
1B, an image processing device 2, and a monitor television 3. The visual recognition device of the illustrated example is for inspecting the faintness or lack of printing of the character A printed on the surface of the object 4, but the visual recognition device of the present invention is not limited to this. It can also be used for inspection of the type, shape, etc.

The objects 4 are arranged at regular intervals on a belt conveyor 5 and are sequentially conveyed to a predetermined inspection position along a conveyance path. An illumination device (not shown) and each of the television cameras 1A and 1B are arranged downward above the inspection position, and each of the television cameras 1A and 1B simultaneously captures an image of the object 4 and outputs a grayscale image signal at the time of inspection. Generate.

The second television camera 1B has a higher resolution than the first television camera 1A, and the first television camera 1A is set to have a wide observation field so as to include the entire object 4; The second television camera 1B is set to have a narrow observation field of view enlarged to include only the portion of interest of the object 4, that is, the portion of the character A in this embodiment. In the vicinity of the inspection position, an object detection sensor such as a photoelectric sensor for detecting that the object 4 has reached the inspection position is provided.

The first and second television cameras 1A and 1B and the object detection sensor are connected to the image processing device 2, and the grayscale image signals obtained by each of the television cameras 1A and 1B are taken into the image processing device 2. Predetermined measurement and recognition processing is performed.

FIG. 2 shows an example of a circuit configuration of the image processing apparatus 2. In this image processing device 2, each A / D converter 6A, 6B digitizes a grayscale image signal from each of the television cameras 1A, 1B and outputs it to the display control unit 7 and the respective binarizing circuits 8A, 8B. . Each binarizing circuit 8A,
8B performs a binarization process on the digital image signal with a predetermined binarization threshold, and stores the binary image in the image memory 9 or outputs the binary image to the measurement unit 12. The other image memory 10 is for storing the grayscale image from the first television camera 1A.

One measuring unit 11 is a first television camera 1
A density average value described later is measured for the grayscale image from A,
The other measuring unit 12 measures the feature values such as the area, the center of gravity, and the principal axis angle of the binary image together with the CPU 13. CP
U13 is a main body of control and calculation, and a ROM 14 for storing programs and a RAM 1 for storing various data.
5 constitutes a microcomputer. RAM1
5 includes, in addition to a reference value for recognition, coordinates of an image obtained by the first television camera 1A and the second television camera 1A.
A dimension ratio k (described later in detail) with respect to the coordinates of an image obtained by B is stored.

The three window memories 16, 17, 18
Among them, the first window memory 16 stores window data for designating the measurement area of the feature amount with respect to the binary image by the first television camera 1A, and the second window memory 17 stores the first television camera 1A. The window data for specifying the measurement area of the average density value for the grayscale image of 1A, and the third window memory 18 for specifying the measurement area of the feature amount for the binary image of the second television camera 1B. Window data is CPU1
3 is written.

FIG. 3 shows a binary image 20 obtained by the first television camera 1A. In this binary image 20, a rectangular window W1 for designating a measurement area of a feature amount is set. ing. In the figure, reference numeral 21 denotes an image portion of the object 4, and the window W1 corresponds to the image portion 21.
Is provided over almost the entire screen so as to include a margin.

FIG. 4 shows a gray-scale image 22 obtained by the first television camera 1A. In this gray-scale image 22, a rectangular window W2 for designating a measurement area of the average density value is set. I have. In the figure, reference numeral 23 denotes an image portion of the object 4, reference numeral 24 denotes a background image portion, and the window W <b> 2 is provided in the background image portion 24 which is sufficiently far away from the image portion 23 of the object 4.

FIG. 5 shows a binary image 25 obtained by the second television camera 1B. In this binary image 25, a rectangular window W3 for designating a measurement area of a feature amount is set. ing. In the figure, reference numeral 26 denotes an image portion of the character A printed on the surface of the object 4, and the window W3 is provided in an appropriate size so as to include the image portion 26.

Returning to FIG. 2, the display control section 7 selects one of a grayscale image signal, a binary image signal, and window data to generate display data. The display data is converted into an analog signal by the D / A converter 19 and output to the video monitor 3.

FIG. 6 shows a control procedure of the CPU 13 for a setting step performed prior to the inspection step. First, at the start of the drawing, the object 4 as a non-defective sample is positioned and arranged at the observation position.
In step ST <b> 1, the CPU 13 causes the image processing device 2 to output an image of the object 4 captured by the first television camera 1 </ b> A. After the grayscale image signal is digitized by the A / D converter 6A of the image processing device 2, the grayscale image is stored in the image memory 10, and the binary image output from the binarization circuit 8A is stored in the image memory 9. Each is stored (see FIGS. 3 and 4).

Next, the CPU 13 sets the first window W1 for the binary image 20, and measures the position G0 of the center of gravity of the image portion 21 of the object together with the measuring section 12 (steps 2, 3). In the next step 4, the CPU 13 performs calibration to calculate a calibration value of the first television camera 1A, that is, a dimension per pixel.

Next, the CPU 13 sets a second window W2 for the grayscale image 22, causes the measuring unit 11 to measure the average density A0 of the pixels included in the window W2, and stores the measured value in the RAM 15. It is registered (steps 5 and 6).

In the next step 7, the CPU 13 sets the character A of the object 4 imaged by the second television camera 1B.
The image in which the portion is enlarged is output to the image processing device 2. The grayscale image signal is digitized by the A / D converter 6B of the image processing device 2, binarized by the binarization circuit 8B, and stored in the image memory 9 (see FIG. 5).

Next, the CPU 13 sets a third window W3 for the binary image 25, measures the area S0 of the image portion 26 of the character A together with the measuring section 12, and uses the measured value as a reference for discrimination. The value is registered in the RAM 15 (steps 8 and 9). Next, the CPU 13 performs a calibration to calculate a calibration value of the second television camera 1B, that is, a dimension per pixel, and then a calibration value of the first television camera 1A and the second television camera 1B. The relationship (dimension ratio k) with the calibration value of 1B is calculated and stored in the RAM 15 (steps 10 and 11). For example, when the calibration value (dimension per pixel) of the first television camera 1A is 1 mm / pixel and the calibration value of the second television camera 1B is 0.2 mm / pixel, the dimension ratio k is 5 : 1.

FIG. 7 shows a control procedure of the CPU 13 for an inspection process performed after the completion of the setting process, and FIG. 8 shows a time chart in the inspection process. When the object 4 to be inspected is conveyed to the observation position by the belt conveyor 5 at the start point in FIG. 7, the object detection sensor outputs a detection signal, and the CPU 13 uses the input of the detection signal as a measurement trigger to execute the first detection. , And the image of the object 4 simultaneously captured by the second television cameras 1A and 1B is output to the image processing apparatus 2.

The grayscale image signals Vd1 and Vd2 for one frame of these images are supplied to the A / D converters 6A and 6A of the image processing apparatus 2, respectively.
6B, the grayscale image signal from the first television camera 1A is sent to one of the measuring units 11
The binary image signal that has passed through the binarization circuit 8A is output to the other measurement unit 12
Are output to the second television camera 1B, and are subjected to predetermined measurement processing.
The value image signal is output to and stored in the image memory 9 (steps 1 to 3).

In steps 2 and 3, the CPU 13 applies the first image to the binary image 20 from the first television camera 1A.
After setting the window W1 and measuring the center of gravity G1 of the image portion (indicated by a dotted line in FIG. 3) of the object together with the measuring unit 12, the difference from the center of gravity G0 is calculated as the positional deviation amount Δ
X _G , ΔY _G , and furthermore, this positional deviation amount is converted into the positional deviation amounts X _W , Y _W (see FIG. 5) of the image obtained from the second television camera 1B using the dimensional ratio k. . For example, if the dimensional ratio k is 5: 1, the positional shift amounts X _W and Y _W are X _W = 5 × ΔX _G and Y _W = 5 × ΔY _G.

In the same steps 2 and 3, the CPU 13 sets the first television camera 1A in parallel with the above-described displacement measurement.
A second window W2 is set for the grayscale image 22 from the image data, and the measuring unit 11 measures the average density value A1 of the pixels included in the window W2, and then illuminates the difference with the average density value A0. Density change amount ΔA due to fluctuation etc.
(= A0-A1).

After the displacement and the density change are measured in this way, the CPU 13 moves the set position of the third window W3 by the displacements X _W and Y _W during the vertical blanking period. , Binarization circuit 8
The binarization threshold values of A and 8B are corrected by the density change amount ΔA (step 4).

Next, during the next frame period, the CPU 13 outputs the binary image 25 from the second television camera 1B.
, A third window W3 after the movement is set, the area S1 of the image portion 26 of the character A is measured together with the measuring section 12, and the measured value is compared with the reference value S1 to make the character A blurred or missing. Is determined (steps 5 and 6). The same procedure is performed on the objects 4 that are successively conveyed, and the measurement processing and the determination processing are performed. When the processing of all the objects 4 is completed, the determination in step 7 is “YE”.
In step S ", the inspection process is completed.

[0029]

According to the present invention, as described above, the amount of displacement of an object is measured from an image from the first image pickup means having a wide observation field of view, and the second image pickup means having high resolution is obtained from the measured value. Since the position of the window is changed by converting the amount of positional deviation of the image and performing a predetermined measurement process, even if the position or orientation of the object changes, the measurement accuracy is not reduced, High-precision inspection becomes possible.
In addition, a correction value of a parameter for image processing is calculated based on the image obtained by the first imaging unit, and the second value is calculated based on the correction value.
Since the image obtained by the imaging means is processed, stable measurement can be performed even if there is a variation in illumination and the like, and remarkable effects such as a reduction in the frequency of inspection errors can be achieved.

[Brief description of the drawings]

FIG. 1 is a perspective view illustrating an appearance of a visual recognition device according to an embodiment of the present invention.

FIG. 2 is a block diagram illustrating a circuit configuration example of the image processing apparatus.

FIG. 3 is an explanatory diagram showing a binary image from a first television camera.

FIG. 4 is an explanatory diagram showing a grayscale image from a first television camera.

FIG. 5 is an explanatory diagram showing a binary image from a second television camera.

FIG. 6 is a flowchart illustrating a control procedure of a setting step.

FIG. 7 is a flowchart showing a control procedure of an inspection process.

FIG. 8 is a time chart showing a temporal flow in an inspection process.

[Explanation of symbols]

 1A, 1B TV camera 2 Image processing device 8A, 8B Binarization circuit 9, 10 Image memory 11, 12 Measuring unit 13 CPU 16, 17, 18 Window memory

Claims (1)

(57) [Claims] 1. A first imaging means capable of including an entire object in an observation field of view, and a resolution higher than that of the first imaging means capable of including only a focused portion of the object in the observation field. A second imaging unit, a displacement calculating unit that calculates a displacement amount of the image of the target object obtained by the first imaging unit, and a second imaging unit based on the image obtained by the first imaging unit. Correction value calculating means for calculating a correction value of a parameter for processing an image obtained by the method, and a dimensional ratio between coordinates of an image obtained by the first image pickup means and coordinates of an image obtained by the second image pickup means A position shift conversion unit that converts the position shift amount calculated by the position shift calculation unit into a position shift amount of the image of the object obtained by the second imaging unit using The corrected image by the correction value calculating means. An image processing means for processing based on the output correction value; and, for an image obtained by the second imaging means, changing the set position of the measurement area by a positional deviation amount converted by the positional deviation converting means, and A visual recognition device comprising: a measurement processing unit that performs a measurement process.