JPH0658741A - Detection method for part inclination - Google Patents

Abstract

PURPOSE:To enable exactly detecting the part inclination regardless of the silhouette state by producing the angle distribution of lines connecting two points apart for specific number of each pixell constituting the image outline. CONSTITUTION:A CCD camera takes picture of a part 4 adsorbed by a adsorption nozzle 2 of a surface mounting machine 1 through a mirror 5 and a lens 6. The image is stored in a frame memory 8 and processed with a high speed processor 11. The part outline is extracted at first in the image processing and straight lines connecting two points apart for a specific number of each of a plurality of pixells constituting the outline are produced. The angle of the straight line from a determined reference line is calculated. By calculating the angle for all straight lines, angle number distribution is produced. Smoothing processing is performed for this angle number distribution until an angle with an angle number over a predetermined fraction in the peak angle number decreases below a predetermined number. The part inclination angle is known from the angle over a predetermined fraction in the peak angle number.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a method for detecting the state of a component mounted on a circuit board, particularly the inclination.

[0002]

2. Description of the Related Art In a surface mounter that conveys and mounts a component on a circuit board, when the component is picked up by a suction nozzle in order to place the component in a correct position at a predetermined position on the board, It is necessary to check that the parts are in the correct position. Therefore, an image of the component is picked up by an appropriate image pickup means to obtain a binary image, and the obtained binary image is subjected to image processing to detect the positional deviation or inclination of the component. , It is corrected so that the posture is correct.

Conventionally, as a method for detecting the inclination of the component as described above, a method for detecting the angle of the main axis of the image of the object (longitudinal direction of the image) is known. This is a method of obtaining an inclination (rotation angle) θ of a binarized image of an object (component) from a predetermined reference line, as shown in FIG. That is, the binary image is represented by two-dimensional coordinates (i, j), and the function representing the shape of the binarized component is f (i, j), the component f.
If the coordinates of the center of gravity of (i, j) are (i _G , j _G ), and the central moment around the (p + q) th center of gravity is M _pq , Therefore, the tilt angle θ is

[0004]

[Equation 1]

[0005] is required.

[0006]

However, in recent years,
Surface mount components (chip components) are becoming smaller, and when the chip components are sucked by the suction nozzle of the surface mounter in order to install the chip components on the circuit board, as shown in FIG. The chip component 13 is not always normally suctioned by the suction nozzle 14, but the suction nozzle 14 may protrude from the component 13 as shown in FIG.

Further, since the chip component vibrates together with the tape when it is transported by the transport tape in the surface mounter, it is not possible to specify in which direction the suction nozzle 14 protrudes from the chip component.

FIG. 12 shows a binary image (silhouette) of the chip component sucked by the suction nozzle of the surface mounter. In the same figure (a), the component is normally adsorbed, but in the same figure (b), the nozzle image 14 'protrudes from the component image 13'. At this time, according to the conventional method, there is a possibility that the position of the spindle 15 of the chip component may be calculated while being shifted to the protruding side of the nozzle image 14 '. Furthermore, since it is not possible to specify in which direction the nozzle protrudes, it is not possible to specify the direction in which the main shaft 15 is displaced. Further, even when the shape of the chip component has irregularities, the same problem may occur because irregularities appear in the silhouette.

Therefore, in such a case, it is difficult to detect the original inclination of the chip component, and the position of the chip component cannot be easily corrected.

Therefore, an object of the present invention is to provide a method for correctly detecting the inclination of a component regardless of the state of the silhouette of the component.

[0011]

According to the present invention, there is provided a method for detecting an inclination of a component by picking up an image including a component installed on a circuit board to obtain a grayscale image and performing arithmetic processing on the grayscale image. , (A) extracting a contour from the grayscale image,
(B) For each of the plurality of pixels forming the contour, a straight line connecting two pixels separated from the pixel by a predetermined number of pixels on the contour is generated, and (c) generated for all of the plurality of pixels. Obtain the angle between the straight line and the predetermined reference line,
A frequency distribution of those angles is generated, and (d) the frequency distribution is smoothed until an angle at which the frequency is equal to or higher than a predetermined ratio of peak frequencies on the frequency distribution is equal to or lower than a predetermined number, (e)
It is characterized in that the inclination angle of the component is detected from an angle having a frequency equal to or higher than a predetermined ratio of the peak frequency on the smoothed frequency distribution.

[0012] In a preferred aspect of the present invention, in (e), an average value of angles having a frequency equal to or higher than a predetermined ratio of the peak frequency is calculated, and the average value is set as an inclination angle of the component.

[0013]

First, the contour of a component is extracted from a binary image obtained by capturing an image including the component. A conventionally known method is used for extracting the contour.

A straight line connecting two pixels separated by a predetermined number from each of the plurality of pixels forming the contour is generated.

An angle formed by a predetermined reference line is calculated from the generated straight line, and the frequency distribution is generated by obtaining this angle for all the straight lines.

With respect to the generated frequency distribution, smoothing processing is performed until the angle at which the frequency is equal to or higher than a predetermined ratio of peak frequencies is equal to or lower than a predetermined number.

Then, the inclination angle of the component can be obtained from the angle having a frequency equal to or higher than a predetermined ratio of the peak frequency on the smoothed frequency distribution.

For example, the angle having the peak frequency may be used as the inclination angle of the component as it is, but the inclination angle of the component can be obtained by obtaining the average value of the angles having the frequencies higher than a predetermined ratio of the peak frequency. .

[0019]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENT FIG. 1 shows a system configuration for implementing the method of the present invention. This system is a CCD for picking up an image of a chip component 4 sucked by a suction nozzle 2 of a surface mounter 1 equipped with epi-illumination 3 via a mirror 5 and a lens 6.
A frame memory 8 that includes a camera 7 and stores an image sent from the camera 7, and the frame memory 8
Image processing high-speed processor 11 that performs a process of taking in the image data stored in the memory via the data bus 10, a personal computer 12 as a controller that controls the surface mounter 1 based on the processed data, a processing operation of the processor 11 and image data A black-and-white monitor 9 for displaying is displayed.

FIG. 2 is a flow chart of the method of the present invention implemented in the processor 11. The flow chart will be described below.

First, in step ST1, the outline of the silhouette of the chip component 4 captured by the CCD camera 7 and obtained as shown in FIG. 3A is traced, and an image as shown in FIG. 3B is obtained. obtain. The contour tracing is performed as follows.

As shown in FIG. 4, the image is scanned in the right direction from the trace starting point S which is appropriately determined, and the point A which first hits the binary image is used as the starting point. And from this starting point A 2
Trace around the value image, and when the point returns to the starting point A, the trace is completed. In the case of the embodiment, the pixels forming the contour are numbered in the order of tracing.

Alternatively, the brightness of the image may be differentiated to obtain a gradient, and the edges constituting the contour may be obtained and traced from the direction of the gradient. In this case, since the pixels forming the contour are not numbered, they need to be numbered separately.

FIG. 5 shows an enlarged portion C of the contour traced as shown in FIG. Here, the pixels forming the contour are referred to as contour points.

Next, in step ST2, a straight line is generated by connecting the contour points, and the straight line and a predetermined reference line (for example, one of the two-dimensional coordinate axes for representing the position of the chip component).
Angle) with respect to all generated straight lines.

That is, as shown in FIG. 5, coordinates (X
2 located at the j-th position from the contour point i indicated by _i , Y _i )
A straight line is generated by connecting two contour points (X _{i + j} , Y _{i + j} ) and (X _ij , Y _ij ). Here, the angle θ _i formed by the obtained straight line and the reference line (coordinate axis X) is calculated by the following equation.

[0027]

[Equation 2]

For example, when j = 5, the straight line shown in FIG. 5 is generated for the contour point i.

When the number of all the contour points forming the contour is n, for the contour point i (= 0, 1, ..., N-1),
Generate a straight line as above. With respect to the straight line generated in this way, θ _i can be sequentially calculated by the above equation (3).

At step ST3, a frequency distribution (histogram) in the range of -179 ° to + 180 ° is generated by plotting the angle obtained at step ST2 on a plane having the horizontal axis as the horizontal axis and the frequency as the vertical axis. .

FIG. 6 shows an example of the histogram generated as described above. In this histogram, since the shape of the silhouette of the target chip component is a quadrangle, the angles obtained from the contour points of the chip component form peaks at 90 ° intervals on the histogram. That is, the original inclination angle (bending angle) of the chip component is Δ (>
0), the peak is −180 ° + Δ, −90 ° + Δ,
Appears at an angle of Δ, 90 ° + Δ.

On the other hand, when the suction nozzle protrudes and appears in the silhouette, since the shape of the nozzle is not generally quadrangle, the contour is not a straight line, and the angle obtained from the contour point is on the histogram. It does not appear as a clear peak. The same applies when the chip component has irregularities.

Therefore, the above histogram is set to ± 45 ° and ±
If you turn back at 135 ° and turn it on top of each other,
The peaks of chip parts distributed every 90 ° will be more clearly captured.

In step ST4, the above-mentioned folding operation is performed to generate a histogram in the range of the angle of -45 ° to + 44 °. FIG. 7 shows the histogram thus generated.

In step ST5, the histogram obtained in step ST4 is subjected to smoothing (noise removal) using a Gaussian function described later, and the peak of the angle calculated from the contour points of the chip part is captured more clearly. To be done.

At step ST6, the smoothness of the histogram smoothed at step ST5 is judged.
That is, as shown in FIG. 8, when the number A of the peak (in this case, maximum) frequency is the threshold value and the number N of the angles Δ _k having the higher frequency is less than or equal to M, smoothing is performed. It is determined that the smoothing has been sufficiently performed, and the process proceeds to step ST7. When N> M, the smoothing is considered insufficient and the process proceeds to step ST8. Here, M is the number (threshold value) appropriately set in the range of 5 to 10 according to the situation of the chip component to be inspected.

[0037] In step ST7, frequency n _k of the angle Δ _k
Then, the average angle is calculated as follows.

[0038]

[Equation 3]

This average angle is the inclination angle Δ of the chip component 4'shown by the silhouette shown in FIG. 9, that is, the inclination angle of the inspection object component.

Depending on the condition of the component to be inspected, the angle itself which is the peak frequency can be adopted as the inclination angle of the component.

On the other hand, in step ST8, the number of repetitions of the smoothing process performed in step ST5 is judged. That is, this number is the predetermined number B
If it has not reached, the process returns to step ST5 to perform smoothing again, and if it has reached, detection of the tilt angle of the target component is stopped (step ST9), and the processing procedure ends.

Here, B is a threshold value set in accordance with the type of the inspection target part, and in the embodiment, the inclination detection procedure is terminated when B = 10 is reached. For example, in the case of a part having a non-rectangular shape, a clear peak is unlikely to be formed when the histogram is generated, and thus a peak satisfying N ≦ M is unlikely to be generated even if the smoothing is repeated.
In such a case, it will be subject to censoring.

Next, the arithmetic processing for executing the above method will be described in detail.

First, the angle θ calculated in step ST2
Let _i be x and the histogram generated for the angle x be f (x).

In the case of the embodiment, since the shape of the parts is a quadrangle, f (x) has peaks at 90 ° intervals. Therefore, f
Histogram h (x) obtained by folding (x) every 90 ° [−45 <
x ≦ 45] is calculated. This is expressed by the following equation.

[0046]

[Equation 4]

Furthermore, h (x) is

[0048]

[Equation 5]

And normalize. This Gh (x) is standard deviation σ
Gaussian function of

[0050]

[Equation 6]

Filter with. That is, the filtering result

[0052]

[Equation 7]

This H (x, σ) is expressed as

[0054]

[Equation 8]

Normalize as follows. Fourier transform of the Gaussian function of equation (7) gives

[0056]

[Equation 9]

Thus, 99.87% of full-width frequencies are | w |
It is within 3 / σ. Therefore, with the Gaussian function g (x, σ)
By filtering Gh (x), the high frequency component (noise) of Gh (x) is removed and Gh (x) is smoothed.

If σ is properly selected, when the component is not inclined, the component has many horizontal and vertical components, so x
When it is inclined by Δ by = 0, a peak above the noise level at x = Δ appears at GH (x, σ).

Since it is difficult to predict an appropriate σ in advance, a Gaussian function g (x, 1) with σ = 1 is obtained as follows.
Repeated smoothing is performed until the number of angles equal to or higher than the noise level (threshold value A%) of the filtering output becomes M or less.

[0060]

[Equation 10]

In the above equation, H ⁽ⁿ⁾ (x) is the nth filtering result, and GH ⁽ⁿ⁾ (x) is the normalization of H ⁽ⁿ⁾ (x) within the range of 0 to 1.

When the number of repetitions n reaches the threshold value B, but the number of angles does not become M or less, the inclination detection of the chip component to be inspected is terminated.

In the above-mentioned embodiment, the tilt of the chip component when it is mounted by the surface mounter is detected, but the method of the present invention is not limited to this, and a component that can be grasped as a binary image. If so, it can be applied to the inclination detection. For example, even if the component is mounted on the circuit board, if it can be captured as a binary image, the inclination can be detected.

[0064]

As described above, when the method of the present invention is used,
Even when the suction nozzle of the surface mounter protrudes from the chip component to be inspected, the inclination of the chip component can be correctly detected, so that the inclination of the component can be correctly corrected regardless of whether or not the nozzle protrudes.

Further, the inclination can be detected even when the chip component has irregularities.

Furthermore, since the method of the present invention detects the inclination from the contour, it can be applied to all parts that can be captured as a binary image.

[Brief description of drawings]

FIG. 1 is a system configuration diagram for implementing a method of the present invention.

FIG. 2 is a flowchart showing a processing procedure of a method executed in the system of FIG.

FIG. 3 is a silhouette of a chip part to be inspected.

FIG. 4 is a diagram illustrating tracing of a contour of a binarized image.

FIG. 5 is a diagram showing the contour portion of the silhouette of FIG. 3 with its constituent pixels.

6 is a diagram showing a frequency distribution of inclination angles of a straight line generated for each pixel in FIG.

7 is a diagram showing a histogram of an angle range of ± 45 ° generated from the frequency distribution of FIG.

FIG. 8 is a diagram showing a histogram obtained by smoothing the histogram of FIG.

FIG. 9 is a diagram showing a tilt angle of a silhouetted chip component.

FIG. 10 is a diagram showing a tilt angle of a binarized object on a coordinate plane.

FIG. 11 is a view showing a state of a chip component sucked by a suction nozzle.

FIG. 12 is a silhouette of a chip component sucked by a suction nozzle.

[Explanation of symbols]

1 ... Surface mounter, 2, 2 ', 14 and 14' ... Suction nozzle, 3 ... Epi-illumination, 4, 4 ', 13 and 13' ... Chip parts, 5 ... Mirror, 6 ... Lens, 7 ... CCD camera, 8
... Frame memory, 9 ... Monochrome monitor, 10 ... Data bus, 11 ... High-speed processor for image processing, 12 ... Personal computer, 15 ... Spindle.

Claims (2)

[Claims] 1. A method for detecting an inclination of a component by performing an arithmetic process on the grayscale image by picking up an image including a component installed on a circuit board, and (a) from the grayscale image A contour is extracted, and (b) for each of a plurality of pixels forming the contour, a straight line connecting two pixels separated from the pixel by a predetermined number of pixels on the contour is generated, (c) the plurality of pixels Of the angle between the straight line generated for all of the above and a predetermined reference line, and a frequency distribution of those angles is generated. (D) An angle having a frequency equal to or higher than a predetermined ratio of the peak frequency is predetermined on the frequency distribution. Smoothing the frequency distribution until the frequency becomes equal to or less than (4), and (e) detecting the inclination angle of the component from the angle at which the frequency is equal to or higher than a predetermined ratio of the peak frequency on the smoothed frequency distribution. Method for detecting the tilt of a part. 2. The method according to claim 1, wherein in (e), an average value of angles having a frequency equal to or higher than a predetermined ratio of peak frequencies is calculated, and the calculated value is used as an inclination angle of the component. Part tilt detection method.