JP3166816B2 - Pattern position finding method and apparatus by image recognition - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method and an apparatus for locating a pattern by image recognition, and more particularly, to a method and a pattern locating method for detecting and locating a target pattern formed on a substrate and serving as a reference by image recognition. Related to the device.

[0002]

2. Description of the Related Art For example, a printed wiring board (hereinafter simply referred to as a board) incorporated in an electronic device or the like generally has a conductive wiring pattern on which various electronic components are mounted, formed on the surface of the board. It is. In the manufacture of electronic devices and the like, in order to automatically mount electronic components on the wiring patterns, patterns serving as position references for the wiring patterns are provided separately from the wiring patterns.

[0003] Since the wiring patterns are usually close together in a dense state due to high-density mounting on a substrate, the reference patterns are formed in a narrow area between adjacent wiring patterns around the wiring patterns. Must be formed.

As described above, in order to accurately position and mount an electronic component on a wiring pattern on a substrate, it is necessary to accurately recognize the position of a reference pattern in advance. Therefore, with the reference pattern as a recognition target, the reference pattern is image-recognized by imaging the substrate surface with an optical camera, and the reference pattern is located based on the recognition data. I have.

[0005]

However, the above-mentioned reference pattern (hereinafter referred to as a target pattern)
Is formed in a narrow area between adjacent wiring patterns (hereinafter, referred to as non-target patterns), so that not only the target patterns but also the non-target patterns are included in the field of view of the camera captured by the optical camera. Will come in. If there is a target pattern and a non-target pattern in the camera field of view in this way, it is necessary to distinguish between them, and if an attempt is made to have a function of excluding the non-target pattern and detecting only the target pattern, There is a problem that the device itself becomes expensive, for example, it is necessary to construct complicated software.

Accordingly, the present invention has been proposed in view of the above problems, and an object of the present invention is to provide a pattern locating system having a function of detecting only a target pattern by a simple process based on image recognition. It is to provide a method and an apparatus.

[0007]

As a technical means for achieving the above object, a method according to the present invention comprises a method of forming a reference target pattern formed on a substrate and a peripheral non-target pattern in a camera field of view. A method for detecting and locating a target pattern formed in a narrow area between adjacent non-target patterns by image recognition, wherein a recognition window excluding the non-target pattern is set in the narrow area. Then, the center of gravity and the area of the target pattern are calculated in the recognition window, it is determined whether the target pattern is in the recognition window based on the area data, and the target pattern is determined based on the center of gravity data based on the determination result. The position is corrected.

The present invention also provides a drive system for moving a substrate on which a target pattern serving as a reference and a peripheral non-target pattern are formed in the X and Y directions, and an optical system for imaging the target pattern and the non-target pattern. In the field of view of the camera, a recognition window for excluding the non-target pattern is set in a narrow area between adjacent non-target patterns, the center of gravity and area of the target pattern are calculated in the recognition window, and the center of gravity and An image recognition system that outputs area data, and a control system that determines whether or not the target pattern is within the recognition window based on the area data and corrects the position of the target pattern based on the center of gravity data based on the determination result. It is characterized by the following.

In the above apparatus, the substrate is long and two target patterns are formed at intervals along the longitudinal direction, and the driving system moves the substrate in the XYθ directions. First and second optical systems for imaging two target patterns and non-target patterns in the vicinity thereof, and excluding the non-target pattern in a narrow area between adjacent non-target patterns in respective camera fields of view. It is possible to include first and second image recognition systems that set a recognition window, calculate the center of gravity and area of the target pattern in the recognition window, and output data of the center of gravity and area.

[0010]

According to the present invention, first, a target pattern formed on a substrate and serving as a reference and a non-target pattern around the target pattern are imaged by an optical system, and a narrowing between adjacent non-target patterns is set within a camera field of view set by the optical system. An image recognition system sets a recognition window for excluding the non-target pattern in a small area. The image recognition system calculates the center of gravity and the area of the target pattern in the recognition window, and the control system determines whether or not the target pattern is in the recognition window based on the area data. The position of the target pattern is corrected by the center of gravity data by moving the substrate in the X and Y directions by the drive system based on the determination result. Thus, only the target pattern can be reliably located by the simple processing based on the image recognition.

[0011]

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described with reference to FIGS.

The positioning device of the present invention has the following schematic configuration. First, in the positioning device, a target pattern (reference pattern) and its surrounding non-target pattern (wiring pattern) are formed on a substrate on which the target pattern in a narrow area between adjacent non-target patterns is recognized. And

As shown in FIG. 3, a schematic configuration of the positioning device includes a driving system 2 for moving the substrate 1 in the XYθ directions;
The field of view a (position, width: see FIG. 1) including the camera 7 having a zoom function of enlarging and capturing an image at a predetermined magnification.
Is an optical system 3 that is set so as to sufficiently capture the target pattern m in consideration of the dimensional accuracy of the substrate 1 and the mounting accuracy on the movable table 6 in accordance with the type of the substrate 1 and the optical system 3 adjacent to the optical system 3 within the camera field of view a. A recognition window b for excluding the non-target pattern n is set in a small area e between the non-target patterns n to be processed, the center of gravity and the area of the target pattern m are calculated in the recognition window b, and the center of gravity and area data are calculated. And a control system 5 that determines whether or not the target pattern m is within the recognition window b based on the area data and corrects the position of the target pattern m with the center of gravity data based on the determination result. Is provided.

More specifically, the drive system 2 includes the control system 5
And a drive mechanism for moving the movable table 6 on which the substrate 1 is positioned and mounted by a predetermined amount in the XY directions based on a drive signal from the controller.

The image recognition system 4 includes a target pattern m and a non-target pattern n based on the image pickup signal from the optical system 3.
Window in which the non-target pattern n is excluded from the camera view a for the image information obtained by capturing the image at a predetermined magnification.
Is set. Then, the center of gravity and the area of the target pattern m in the recognition window b are calculated, and the data of the center of gravity and the area are output to the control system 5. Control system 5
Determines whether the target pattern m is within the recognition window b based on the area data output from the image recognition system 4, outputs a drive signal based on the determination result to the drive system 2, and outputs the drive signal The movable table 6 is actuated based on this, and the position of the target pattern m on the substrate 1 is corrected using the center of gravity data.

The method of locating the target pattern m by image recognition is performed in the following manner based on the camera visual field a shown in FIG. 1 and the flowchart shown in FIG.

First, when the substrate 1 placed on the movable table 6 is at a substantially correct position, the target pattern m and the non-target pattern n are enlarged by the camera 7 of the optical system 3 at a predetermined magnification. An image is taken, and a camera field of view a is set as shown in FIG. The camera field of view a is such that the target pattern m is located substantially at the center, and the size of the field of view is small enough to fit in the field of view a even in consideration of variations in the placement of the substrate 1. Next, a recognition window b excluding the non-target pattern n in the camera view a is set in the image recognition system 4. The recognition window b is selected to be narrow enough that the non-target pattern n does not fall within the recognition window b even if the variation in the placement of the substrate 1 is taken into consideration, and that the target pattern m is always large enough to fall within the recognition window b. Is done.

As described above, the recognition window b is provided with a clearance with the non-target pattern n in the narrow area e and excluding the non-target pattern n around the same, thereby enabling the recognition window b to be described later in the recognition window b. The center of gravity and area data of the target pattern m can be accurately calculated. On the other hand, by setting the recognition window b to be very narrow, the target pattern m may protrude from the recognition window b (see FIG. 1A).

Next, a method for positioning each substrate will be described. The substrate 1 is placed on the movable table 6 and the optical system 3
After inputting an image (step 100), the target pattern m which is partially exposed in the recognition window b
, The center of gravity and the area of the portion of the target pattern m present in the recognition window b are calculated by the image recognition system 4 (step 101). The control system 5 determines whether the target pattern m is within the recognition window b or protrudes based on the area data output from the image recognition system 4 (step 102). The determination as to whether or not the image is within the recognition window b is made by determining whether the area data output from the image recognition system 4 is the target pattern m stored and held in advance.
Is determined based on whether or not it matches the normal area value.

Here, the number of times of calculation of the center of gravity and the area of the target pattern m is set to a predetermined number of times, for example, two (step 1).
03), if the number of calculations is three or more, an alarm is issued by a buzzer, a lamp, or the like, indicating that the position cannot be corrected (step 104). Step 105) is executed.

Therefore, if it is determined that the target pattern m protrudes based on the area data output from the image recognition system 4 (step 102), the number of times of calculation of the center of gravity and the area is one at this time. (Step 103),
A drive signal is output from the control system 5 to the drive system 2, and the drive system 2 operates the movable table 6 in the X and Y directions.
The position of the target pattern m is corrected using the center-of-gravity data (step 105). The operation of the movable table 6 in the XYθ directions may be performed separately or simultaneously in the X direction and the Y direction.

After the image input (step 100) by the optical system 3 is executed again, the center of gravity and the area of the target pattern m are calculated (step 101). Next, if the area data output from the image recognition system 4 matches the above-described normal area value, it means that the target pattern m is within the recognition window b (see FIG. 1B). The position correction based on the center-of-gravity data is executed (step 106), and the movable table 6 is operated in the XY directions by the drive system 2 based on the drive signal output from the control system 5, and the target pattern m
Is arranged at the center of the recognition window b, and the positioning of the target pattern m is completed (see FIG. 1C).

On the other hand, if the area data output from the image recognition system 4 does not match the normal area value, the target pattern m still protrudes from the recognition window b, and at this time the center of gravity and Since the number of times of calculating the area is two, the position correction (step 105) based on the center-of-gravity data is executed as described above. Then, the optical system 3
(Step 100), the center of gravity and area of the target pattern m are calculated (step 101), and it is determined whether or not the target pattern m is within the recognition window b based on the area data (step 102).

If the target pattern m is within the recognition window b, the position is corrected based on the center of gravity data (step 106). If the target pattern m still protrudes from the recognition window b, the center of gravity and Since the number of times of calculating the area is three, an alarm is issued by a buzzer or a lamp (step 104), and the positioning of the target pattern m is stopped.

In the above embodiment, the case where one target pattern m is formed on the substrate 1 has been described. However, as shown in FIG. 4, the substrate 1 is long and extends along its longitudinal direction. The first and second target patterns m,
The present invention is also applicable to a case where m ′ and non-target patterns n and n ′ are formed.

In this case, the apparatus of the present invention comprises first and second optical systems 3 and 3 ′ for imaging two target patterns m and m ′ and non-target patterns n and n ′ around the target patterns m and m ′. Two camera fields of view of the target pattern m, m 'and the non-target pattern n, n' are set, and the target is set in a narrow area between adjacent non-target patterns n, n 'in each camera field of view. First and second image recognition for setting a recognition window for excluding the outer patterns n and n ', calculating the center of gravity and area of the target pattern m and m' within the recognition window, and outputting the center of gravity and area data And systems 4 and 4 ′.

Further, in this case, since there are two target patterns, it is possible to determine the position in the θ direction. Therefore, the control system 15 sends a control signal in the XYθ direction to the drive system 12. In the drive system 12, the movable table 16 is moved in the XYθ directions to perform positioning.

[0028]

According to the present invention, the center of gravity of a target pattern is determined by discrimination based on the area data of the target pattern based on a recognition window set in the camera's field of view where the target pattern and its surrounding non-target patterns are imaged. Since the position correction based on the data is executed, only the target pattern can be reliably located by simple processing based on image recognition, thereby improving the reliability and providing an inexpensive device, and its practical value. Is great.

[Brief description of the drawings]

FIGS. 1A and 1B are views showing a camera field of view for explaining the method of the present invention, wherein FIG. 1A shows a state in which a target pattern protrudes from a recognition window, and FIG. Indicates that you are in the recognition window,
(C) shows a state where the target pattern comes to the center of the recognition window by the second position correction.

FIG. 2 is a flowchart showing the procedure of the method of the present invention.

FIG. 3 is a block diagram showing a schematic configuration of the apparatus of the present invention.

FIG. 4 is a block diagram showing an application example of the present invention.

[Explanation of symbols]

 Reference Signs List 1 substrate 2 drive system 3, 3 'optical system 4, 4' image recognition system 5, 15 control system 6, 16 movable table a camera field of view b recognition window e narrow area m target pattern n non-target pattern

──────────────────────────────────────────────────続 き Continued on the front page (58) Fields surveyed (Int. Cl. ⁷ , DB name) G01B 11/00-11/30 G01N 21/956 H01L 21/68 G06T 1/00 305 H05K 13/04

Claims (2)

(57) [Claims] 1. A target pattern formed in a narrow area between adjacent non-target patterns in an image field of a camera which images a target pattern formed on a substrate and serves as a reference and a non-target pattern around the target pattern. In the method of detecting and positioning by setting, the recognition window excluding the non-target pattern is set in the narrow area, first, the
Includes the target pattern at the approximate center of the recognition window
The area and the center of gravity of the target pattern calculated in the state where
Stored as reference area data and reference centroid data
The target pattern calculated in the recognition window
The area and the center of gravity of the object based on the area data
Determine whether the pattern is within the recognition window and determine
The area data is based on the result and the reference area data.
If they match, the position is corrected using the center of gravity data and the area data is corrected.
If the data does not match the reference area data,
After the position is corrected by the data, the image of the target pattern is
The target pattern by repeating the
A pattern position finding method based on image recognition, wherein the position is corrected to the position of the reference centroid data . 2. A drive system for moving a substrate on which a target pattern serving as a reference and a non-target pattern around the target pattern are formed in the XY directions, an optical system for imaging the target pattern and the non-target pattern, and a camera within its field of view. A recognition window for excluding the non-target pattern is set in a narrow area between adjacent non-target patterns, and a center of gravity and an area of the target pattern calculated in the recognition window are set as a reference.
An image recognition system that stores and outputs the center-of-gravity data and the reference area data, and determines whether or not the target pattern is in a recognition window based on the area data, and based on the determination result, the area data is the reference area data. Match
If the position is corrected by the center of gravity data and the area data is
If they do not match the reference area data,
After the position correction, the image recognition and the position of the target pattern are performed again.
Of the reference centroid Data target pattern by repeating the correction
A pattern position locating device by image recognition, comprising: a control system for correcting a position in a position.