JP3342196B2 - Image recognition device and image recognition method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image recognizing apparatus and an image recognizing method for recognizing an image by converting a video signal representing the brightness of an object picked up by an image pick-up device into a digital signal.

[0002]

2. Description of the Related Art This kind of image recognition apparatus is disclosed in Japanese Patent Laid-Open No. 4-137.
No. 599, a board positioning mark of a printed circuit board, which is a subject, is imaged by a board recognition camera as an imaging device, and a recognition circuit recognizes and processes the imaged screen to recognize the position of the board positioning mark. The mounting position of the electronic component is corrected based on the recognition result.

In this type of image recognition device, the image pickup device outputs the intensity of brightness of each pixel of the image of the picked-up subject as a video signal, and the video signal is an analog signal. It is converted into a digital signal and input, and the digital signal is subjected to recognition processing. When converting into a digital signal, a predetermined range of brightness intensity values is divided into, for example, 0 to 255 and converted into digital values. Is called a level, a width from the minimum value to the maximum value to be digitally converted is called a gain, and A / D conversion is performed by determining these two parameters.

[0004]

However, in the above prior art, the A / D is determined by the way of taking these two parameters.
Depending on how to select the range of brightness intensity to be converted, there are cases where the same video signal captured by the imaging device can and cannot be recognized even if the same video signal is captured. However, there are cases where recognition processing can and cannot be performed when the subject is different, and it is difficult for the operator to determine the optimum value of this parameter.

Accordingly, an object of the present invention is to make it possible to obtain an optimum range of the intensity of a video signal to be converted when A / D converting a video signal picked up by an image pickup apparatus.

[0006]

SUMMARY OF THE INVENTION The present invention For this purpose, an image recognition apparatus that performs recognition processing of the image by converting the video signal into a digital signal representative of the brightness of the subject captured in the image pickup device, a digital video signal Set to convert to signal
The minimum or maximum brightness of the video signal to be converted
Recognition processing is performed by changing the large value by a predetermined amount, and recognition processing is performed.
A determination means for automatically determining an intensity range from the minimum value to the maximum value of the video signal for converting the video signal into a digital signal based on a result is provided.

[0007] The present invention is an image recognition apparatus that performs recognition processing of the image by converting the video signal into a digital signal representing the brightness of the imaged object of the imaging device, de video signal
Of the video signal set to convert to digital signals
Change the minimum value of the brightness to be converted to a large value by a predetermined amount
And the brightness of the set video signal to be converted
A changing means for changing the maximum value to a large value by a predetermined amount , and converting the minimum value and the maximum value changed by the changing means into a digital signal to perform an actual recognition process, whereby the recognition process is successful.
There is provided determining means for determining a range from the minimum value to the maximum value as an intensity range of a video signal to be converted .

[0008] The present invention is an image recognition apparatus that performs recognition processing of the image by converting the video signal into a digital signal representing the brightness of the imaged object of the imaging device, de video signal
Of the video signal set to convert to digital signals
Change the minimum or maximum brightness to be converted by a predetermined amount
Performing a recognition process, and converting the video signal to the minimum value of the brightness of the video signal to be converted into a digital signal based on the recognition process result .
A search means for actual recognition processing performed recognition process by changing the intensity range to Luo maximum find a set of intensity range to be successful, the strength recognizable process finds that the search means
From the set of intensity ranges, for intensity ranges where recognition processing is not possible
Deciding means for deciding the strength range which has the most margin .

[0009] The present invention is an image recognition method which converts the video signal into a digital signal representative of the brightness of the subject captured in the image capturing apparatus performs the recognition processing of the image, de video signal
Of the video signal set to convert to digital signals
Change the minimum or maximum brightness to be converted by a predetermined amount
And perform recognition processing.
A determination step is provided for automatically determining an intensity range from the minimum value to the maximum value of the brightness of the video signal for converting an image signal into a digital signal.

[0010]

According to the first aspect of the present invention, the determining means includes a video signal.
Which converts the video signal into a digital signal.
Automatically determine the intensity range up to. According to the configuration of the second aspect, it is set to convert the video signal into a digital signal.
The minimum brightness of the video signal to be converted
Video signal that has been changed by
The maximum value of the brightness to be converted to a large value by a predetermined amount.
The changing means converts the changed intensity range of the video signal into a digital signal and performs actual recognition processing, and the determining means determines an intensity range in which the recognition processing is successful as an intensity range to be converted.

According to the third aspect of the present invention, the search means includes a
The screen that was set to convert the signal to a digital signal
Predetermined minimum or maximum brightness of image signal to be converted
The actual recognition process is changed and the recognition process succeeds.
A set of intensity ranges was found, and search means found them
The determination means recognizes from the set of intensity ranges that can be recognized.
An intensity range that has the most allowance for an intensity range that cannot be processed is determined. With the fourth feature, at decision step automatically it determines the intensity range of the maximum value from the brightness minimum value of the video signal for converting the video signal into a digital signal.

[0012]

BRIEF DESCRIPTION OF THE DRAWINGS FIG. 2 to 4, reference numeral 1 denotes an XY table which moves in the X and Y directions by the rotation of an X-axis motor 2 and a Y-axis motor 3, on which a chip-shaped electronic component 4 (hereinafter referred to as a chip component or component) is mounted. The printed circuit board 5 to be mounted is placed.

Reference numeral 6 denotes a supply table on which a number of component supply devices 7 for supplying the chip components 4 are provided. Reference numeral 8 denotes a supply table drive motor, which rotates the ball screw 9 so that the supply table 6 is guided by the linear guide 10 via a nut (not shown) fitted to the ball screw 9 and fixed to the supply table 6. Moving. Reference numeral 11 denotes a rotary table which rotates intermittently.
The mounting heads 13 having four 2 are arranged at regular intervals in accordance with the intermittent pitch.

The stop position of the mounting head 13 where the suction nozzle 12 descends and sucks and picks up the component 4 from the supply device 7 is the suction station, and the suction nozzle 12 located at the outermost position of the rotary table 11 at the suction station. The part 4 is sucked. The position at which the mounting head 13 stops next is the angle correction station, and the suction nozzle 12 is rotated in the θ direction by the nozzle rotation roller 15 based on the recognition result by the component recognition camera 14, and the position of the rotation angle of the component 4 The displacement is corrected.

Reference numeral 18 denotes a base, and the XY table 1 moves XY with respect to the base 18. Reference numeral 19 denotes a board recognition camera, and a support plate 2 fixed to the base 18.
The image of the printed circuit board 5 fixed to 0 and placed on the XY table 1 is taken. The printed board 5 is provided with a board recognition mark 22 as shown in FIG.
The recognition camera 19 takes an image of the reflected light from the mark 22 which has been appropriately illuminated, and the position of the substrate 5 on the XY table 1 is recognized. As shown in FIG. 4, the marks 22 are usually provided one by one at opposing corners (corners) of the substrate 5.
Individually attached. The mark 22 shown in FIG. 4 is circular, but may be another shape such as a triangle, a square,
Further, a part of the pattern on the substrate 5 may be used.

Reference numeral 23 denotes a supply conveyor which conveys the printed circuit board 5 and supplies it onto the XY table 1. Next, a control circuit of the electronic component automatic mounting apparatus will be described with reference to FIG. In FIG. 1, reference numeral 25 denotes a CPU, based on various data stored in a RAM 26, in accordance with a program stored in an R0M 28, a mounting operation of a chip component 4 on a printed circuit board 5 of an electronic component mounting apparatus, and a transfer of the substrate 5. And various operations related to operations such as recognition of the substrate 5 or the component 4.

A drive circuit 31 and a recognition processing device 32 are connected to the CPU 25 via an interface 30. The X-axis motor 2 and the Y-axis motor 3 are driven by the drive circuit 31, and the recognition processing device 32 performs recognition processing of the image of the recognition mark 22 captured by the board recognition camera 19. Also,
The recognition processing device 32 also performs processing for recognizing the captured image of the chip component 4 by the component recognition camera 14.
A / D conversion unit 34 that converts a video signal, which is an analog signal based on the brightness of an image captured at the same time, into a digital signal
And a recognition processing unit 35 for recognizing and processing the digital signal.

The CPU 25 outputs an instruction to the recognition processing device 32 to perform recognition processing on the images captured by the imaging cameras 14 and 19, and receives the result of the recognition processing from the recognition processing device 32. The video signal is, for example, a waveform in which the luminance changes along the vertical axis in the scanning direction of the scanning line of the camera, that is, the horizontal axis as the time axis, as shown in FIG. A minimum value and a maximum value to be converted are determined, and the brightness intensity range between the minimum value and the maximum value is divided into 256 equal parts, and is converted into a digital value for each pixel of the image. The bottom solid line in FIG. 5 is the position where the brightness captured by the camera 19 is “0”, and the position above the solid line is the minimum value that is converted to “0” by digital conversion. Is called a level, and the position of the solid line at the top is converted into a digital value to “256”.
And the interval between the minimum value and the maximum value is referred to as a gain. The level and the gain, which indicate the intensity range of the brightness of the video signal to be converted into the digital signal, are stored in the level memory 36 and the gain memory 37 provided in the A / D conversion unit as A / D conversion parameters, respectively. . Even if the setting values of the level and the gain are the same, recognition processing cannot be performed or cannot be performed if the waveform of the video signal is different depending on the imaging object or the state of illumination, etc.
The setting values of the level memory 36 and the gain memory 37 can be appropriately changed by the recognition processing unit 35.

The recognition processing section of the recognition processing device 32 performs the recognition processing of the board recognition mark 22 according to the flowchart of FIG. The operation of the above configuration will be described below. First, an automatic operation is started by a switch (not shown). Next, the printed circuit board 5 is conveyed from an upstream device (apparatus for applying an adhesive) by a supply conveyor 23, transferred onto the XY table 1, and positioned at a predetermined position on the table 1 by a positioning device (not shown). Fixed.

That is, the X-axis motor 2 and the Y-axis motor 3
Driven by the drive circuit 31 under the control of the PU 25, the rotation X
The Y table 1 moves so that one of the board recognition marks 22 of the board 5 is located below the board recognition camera 19. At this time, the theoretical (design) position of the board recognition mark 22 is set at the center of the screen of the camera 19, and the camera 19 is positioned so that the actual mark 22 is not disturbed even if it is shifted. The field of view is secured.

First, the camera 19 picks up the image of the recognition mark 22, but the image as it is is shown in FIG. , And the background of the mark 22 is bright. Next, when the CPU 25 instructs the recognition processing device 32 to perform the recognition processing, the recognition processing unit 35 operates according to the flowchart of the recognition processing in FIG.

That is, first, A / D conversion is performed according to the values stored as initial values in the level memory 36 and gain level 37 of the A / D converter 34, and recognition processing is performed using the digital signal. Select a model to be described later. In this case, the level value as the initial value is set to the position of the bottom solid line in FIG. 5 where the brightness (luminance) of the video signal of the camera 19 is “0”, and the gain as the initial value is a small value. It is assumed that the maximum brightness converted into a digital signal obtained by adding a gain to the level is set to be smaller than the brightness of the minimum video signal.

Next, the recognition processing unit 35 determines the degree of coincidence between the model, which is a circular theoretical mark conforming to the mark 22, and the imaged mark 22 by so-called template matching in which the theoretical mark is moved little by little. The mark detection process is performed by comparison using the method. However, the mark 22 and the background image have the same brightness.
The image cannot be matched, and there is no position that reaches the predetermined matching degree, so that the processing ends abnormally.

Next, the recognition processing section 35 performs an operation of finding an optimum gain and level that can perform the recognition processing. That is,
First, the value of the level stored in the level memory 36 is fixed to the minimum value (the value of “0” which is the position of the bottom solid line in FIG. 5), and the gain of the gain memory 37 is changed from the minimum value to a large value. performs recognition processing according to the flowchart of each time 6 by changing, abnormal termination, i.e. the mark on the theoretical model
And the digital signal converted by the A / D converter 34.
The image of the mark that has been captured
Not, like the value of the level by increasing a predetermined amount if recognition failure
Normal operation, that is, the theoretical mark mode
Dell and the captured image of the mark reach
Then, the minimum value of the level at which recognition is successful is detected.

Next, the level value is set to the minimum value of the successful recognition, the gain value is changed from small to large, and the width of the gain for which the recognition is successful is detected. Thereafter, the level value is increased by a predetermined amount in the same manner. The range of gain values for which recognition is successful is detected. Next, when the level value is increased and a specified amount of gain width cannot be obtained, the level value is set as the maximum level value of successful recognition. When the gain and the level are selected as shown in FIG. 5 as an example in which the recognition is successful, the digitally converted waveform is normalized as shown in FIG. 8, and if this is viewed as an image, as shown in FIG. The change in brightness between the background and the mark is clear. In this case, the value of the degree of coincidence is increased by the template matching, and the recognition processing in FIG. 6 ends normally.

In this way, for example, as shown in FIG. 10, a set of brightness intensity ranges of a video signal to be converted into a digital signal that can be recognized, that is, a level of recognition success and a gain
A recognition success range as a set of combinations with the combination is detected, and a point having the most margin in gain and level in the recognition success range is set as an optimum point, and the level and gain values are stored in the level memory 36 and the gain memory 37. I do. As shown in FIG. 10, the method of finding the optimum point is to find an intermediate value between the minimum value and the maximum value at which the level can be recognized, and to determine and determine an intermediate value of the gain width of the successful recognition at the intermediate value. It is.

Next, a recognition process is performed as described above.
The position of the mark 22 is detected, the amount of displacement from the theoretical position of the mark 22 is calculated by the recognition processing unit 35, and the CPU 25 stores the value in the RAM 26. Next, RA
The suction head 13 moves to the suction station, the suction nozzle 12 moves down, and the motor 8 rotates to a desired position, based on the data indicating the component type and the mounting position in the mounting order (not shown) stored in the M26. The component supply device 7 that supplies the component 5 is located at the suction position, and the chip component 4 is taken out.

Next, the suction nozzle 12 for sucking and holding the chip part 4 is moved to each work station by intermittent rotation of the rotary table 11, and at the recognition station, the suction nozzle 12 for the part 5 is moved by the camera 14 at the recognition station. In the angle correction station, the angle of the component 4 is corrected by the rotation of the nozzle rotation roller 15 based on the recognition result, and reaches the mounting station. At this time, the angle deviation in the θ direction of the substrate 5 already calculated by the substrate recognition is also corrected.

Next, the XY table 1 is used to correct the position of the recognition result at the recognition station and to correct the positional deviation already calculated at the time of board recognition so that a desired mounting position is obtained.
Is moved, and the chip component 4 is mounted on the printed board 5 by the lowering of the suction nozzle 12. Suction nozzle 1 described above
The suction operation of the chip component 4 and the mounting operation on the printed circuit board 5 are performed in the same manner for the suction head 13 subsequent to the second.
Hereinafter, components are mounted on the printed circuit board 5 in the same manner.

Next, when the component mounting on the printed board 5 is completed, the board 5 is discharged from the XY table 1 and the next board is conveyed by the supply conveyor 23 in the same manner as described above and is placed on the table 1. The position is fixed, the values in the level memory 36 and the gain memory 37 are set to the values of the optimum points obtained as described above, and the recognition operation of the two recognition marks 22 is performed in the same manner as described above. Then, the component mounting operation is performed without changing the values of the level memory 36 and the gain memory 37.

In this way, every time the board 5 is changed, board recognition is performed. However, while the recognition is successful, the level and gain values are not changed. In the case of the substrate 5 of the same lot, even if the state of the mark 22 changes slightly, the recognition succeeds with the level and gain values set in this way.
When the lot of the substrate 5 changes and the imaging result of the mark 22 becomes, for example, as shown in FIG. 11 and a part of the mark 22 becomes a dark image, the shape of the mark 22 is recognized as a circle in the conventional setting of the level and the gain. No, the process ends abnormally due to the recognition process of FIG.

In this case, the operation of detecting the optimum point by changing the level and gain again is performed as described above, and the new level and gain are stored in the memories 36 and 37. In such a case as shown in FIG. 11, the video signal along the scanning line at the position indicated by the solid line at the center of the image is as shown in FIG. 12, and in such a case, as shown in FIG. ,
When the level is set, the image is digitally converted as shown in FIG. 13, and the image to be recognized has a contrast between the mark 22 and the background as shown in FIG. 14, and the recognition processing in FIG. You can do it.

In this way, if the new lot has the same tendency as in FIG. 11, the recognition processing can be performed at the same level and gain. If the one shown in FIG. 11 is unique, the operation for detecting the level and gain of the optimum point is performed again on the substrate 5 of the mark 22 as shown in FIG. In addition, based on the first optimum point set as a reference, the optimum point detected when the recognition ends abnormally in the middle is applied only to that substrate, and the next reference level and gain from the next substrate 5 are set. It can be used, or if this way the number of abnormal terminations becomes the set number of times if it ends abnormally many times using the optimal point as the first reference Alternatively, the reference optimal point may be changed. The optimum point to be changed may be determined by, for example, taking an average value from the values detected as the optimum points so far.
May be obtained from the most overlapping portion in the recognizable range shown in FIG.

In this embodiment, the automatic electronic component mounting apparatus for mounting the electronic component 4 on the printed circuit board 5 has been described. However, when the electronic component 4 is mounted on the printed circuit board 5, the electronic component 4 is temporarily fixed. When the position of the substrate 5 fixed on the XY table is recognized by an adhesive application device that applies an adhesive to the printed circuit board 5, and cream solder for soldering electronic components to the printed circuit board 5 is applied to the substrate. 5
The present invention can also be applied to a case where the position is recognized in order to align the substrate 5 with a screen plate in a screen printing machine for printing on a screen.

In this embodiment, the video signal picked up by the component recognition camera 14 uses the same A / D converter as the video signal picked up by the camera 19, but the level when the video signal of the camera 14 is digitally converted is used. , Gain is memory 36,
Instead of using 37, there is another memory and it is set there. Furthermore, the level and gain for digitally converting the video signal captured by the component recognition camera 14 can be automatically set to the optimum values as in the present embodiment, and these settings can be set for each component type. You may make it.

[0036]

As described above, according to the present invention, a video signal
Of the video signal that was set for the brightness of the
Change the minimum or maximum brightness by a predetermined amount to
Of the video signal based on the recognition processing result.
Before the brightness of the video signal to convert the signal to a digital signal
Since the intensity range from the minimum value to the maximum value is automatically determined, the labor of the operator can be saved in determining the optimum intensity range. Further, since the intensity range in which the recognition process is actually performed and succeeds is determined as the intensity range to be converted, the intensity range in which the recognition process succeeds can be reliably obtained.

The recognition process found by the search means is
From the set of possible intensity ranges, the
Since the strength range that has the most room for the surroundings is determined, the optimum strength range is determined.

[Brief description of the drawings]

FIG. 1 is a control block diagram of the present invention.

FIG. 2 is a perspective view of an XY table of the electronic component automatic mounting apparatus of the present invention.

FIG. 3 is a perspective view of the mounting device.

FIG. 4 is a plan view of the mounting device.

FIG. 5 is a diagram showing a video signal waveform.

FIG. 6 is a diagram showing a flowchart of a recognition process.

FIG. 7 is a diagram illustrating a captured image of a board recognition mark.

FIG. 8 is a diagram showing a video signal normalized by digital conversion.

FIG. 9 is a diagram illustrating an image in a state where recognition processing is performed after digital conversion.

FIG. 10 is a diagram showing a graph of a recognition success range in which a level and a gain are set on the vertical and horizontal axes.

FIG. 11 is a diagram illustrating a captured image of a board recognition mark.

FIG. 12 is a diagram showing a video signal waveform.

FIG. 13 is a diagram showing a video signal normalized by digital conversion.

FIG. 14 is a diagram illustrating an image in a state where recognition processing is performed after digital conversion.

[Explanation of symbols]

5 Printed Circuit Board 19 Board Recognition Camera (Imaging Device) 22 Board Recognition Mark (Subject) 32 Recognition Processor 34 A / D Converter 35 Recognition Processor (Determining Means) (Change Means) (Searching Means) 36 Level Memory 37 Gain Memory

──────────────────────────────────────────────────続 き Continuation of front page (51) Int.Cl. ⁷ identification code FI G06T 1/00 G06T 1/00 305C H05K 13/04 H05K 13/04 M (58) Fields investigated (Int.Cl. ⁷ , DB G01N 21/88-21/956 G01B 11/00 G06T 1/00 280 G06T 1/00 305 H05K 13/04

Claims (4)

(57) [Claims] 1. An image recognition apparatus for converting a video signal representing the brightness of a subject imaged by an imaging apparatus into a digital signal and performing image recognition processing, wherein the video signal is converted into a digital signal.
The light to be converted of the video signal set for conversion
Change the minimum or maximum value of the
And converting the video signal into a digital signal based on the recognition processing result from the minimum value to the maximum value of the video signal.
An image recognizing device, further comprising: a determination unit for automatically determining an intensity range up to a value . 2. An image recognition device for converting a video signal representing the brightness of a subject imaged by an imaging device into a digital signal and performing image recognition processing, wherein the video signal is converted into a digital signal.
The light to be converted of the video signal set for conversion
Change the minimum value of Rusa to a large value by a predetermined amount and set
The maximum brightness of the video signal to be converted
Changing means for changing by a predetermined amount to a value, or the minimum value recognition process does the actual recognition process by converting the minimum value and maximum value said changing means is changed to digital signal is successful
An image recognition device provided with a determination means for determining a range from a maximum value to a maximum value as an intensity range of a video signal to be converted . 3. An image recognition device for converting a video signal representing the brightness of a subject imaged by an imaging device into a digital signal and performing image recognition processing, wherein the video signal is converted into a digital signal.
The light to be converted of the video signal set for conversion
Change the minimum or maximum value of the
From the minimum value to the maximum value of the brightness of the video signal to be converted into a digital signal based on the recognition processing result.
Searching means for changing the intensity range of the target and performing actual recognition processing to find a set of intensity ranges in which the recognition processing is successful ;
A set of intensity ranges that can be recognized by the search means
The most margin for the intensity range where recognition processing is not possible.
An image recognition apparatus, comprising: a determination unit that determines an intensity range that can be held. 4. An image recognition method for converting a video signal representing the brightness of a subject imaged by an imaging device into a digital signal and performing image recognition processing, wherein the video signal is converted into a digital signal.
The light to be converted of the video signal set for conversion
Change the minimum or maximum value of the
Performs physical, brightness said minimum value of said video signal for converting the pre SL video signal to digital <br/> barrel signals based on the recognition processing result
An image recognition method, further comprising a determining step of automatically determining an intensity range from to a maximum value .