JPH11132737A - Image processing method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image processing method recognizing the image of a ball of BGA(ball grid array) parts at a high speed and with high accuracy. SOLUTION: Images 3a of balls arranged vertically and laterally on one face of BGA parts are obtained, and vertical direction straight lines and lateral direction straight lines (M1 to M6 , N1 to N6 ) which pass the images 3a of each ball are obtained from the acquired image data of the balls. An inspection area 20 surrounding the images 3a of the balls is set in a region including positions of the balls obtained from the arrangement information of the balls and crossing points of the vertical direction straight lines and lateral direction straight lines, and images of the balls in this set inspection area are processed to determine the central position of each ball and determine the central position and inclination of the BGA parts 1 from the central position. Since an inspection area is not set in a noise 3b, a highly precise image processing is made possible.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing method, and more particularly, to an image processing method for correcting a suction posture of an electronic component and inspecting a shape of the electronic component by using a chip mounter and mounting the electronic component on a substrate.

[0002]

2. Description of the Related Art As electronic components such as integrated circuits, there are known a large number of small spherical ball-shaped objects (hereinafter referred to as balls) arranged on a bottom surface in a two-dimensional array. Such an electronic component is a ball grid array (hereinafter referred to as BGA).
Parts). When mounting a BGA component on a printed circuit board,
The ball is melted by placing the GA component and reflow heating the printed circuit board, and the BGA component is joined to the printed circuit board.

In joining such BGA parts, a conventional Q
The use of an FP (Quad Flat Package) recognition method has the following problems. That is, in the conventional QFP recognition method, as shown by the arrow A in FIG.
A search is performed from the outside of the image field of view to the center direction to detect the position of the lead. However, since the lead 40a always exists on the outer periphery of the QFP component, the edge 40b first found on the search line is: Could be considered part of the lead. However, when this method is used for recognition of BGA components, as shown in FIG.
Bright portion (for example, foreign matter, scratch, etc.) 41a on the bottom surface of 1
Exists, the foreign matter 41 searched along the line B
Since a is recognized as a ball, the first edge 41b to be found may not always be a ball.

In this case, since the actual position of the ball 41c cannot be correctly recognized, and the accurate position and inclination of the BGA component cannot be determined, the BGA component is accurately mounted at a predetermined position. Can not do. For this reason, there is a problem that the recognition of the BGA component using the conventional QFP recognition method is limited to a part of the BGA component which satisfies the condition that all the bottom surfaces are uniformly dark. .

In order to solve this problem, various new methods have been proposed for recognizing images of BGA components. Japanese Patent Application Laid-Open Nos. Hei 5-296724, Hei 6-288732, Hei 7-320062. It is disclosed in gazettes and the like.

In Japanese Patent Application Laid-Open No. Hei 5-296724, the first step of imaging a target component while simultaneously applying transmissive illumination and reflective illumination to obtain a component pattern, and the fact that the background has a high brightness due to the transillumination, the component body A second step of detecting a boundary between the object and the background, a third step of detecting a rough center and inclination of the component from the obtained boundary, and a processing area for electrode detection using the detected center and inclination. Set, analyze the luminance distribution in the processing area, and detect the position of the electrode.
The process is adopted. According to this method, it is possible to easily and accurately set a processing area of the reflection recognition method that enables high-speed recognition by using high-speed tilt and center detection in the transmission recognition method.

On the other hand, in Japanese Patent Application Laid-Open No. 6-288732, an electrode is used to obtain an image pattern of a component arranged in a grid inside a body, and a tilt for detecting a general tilt of the component based on the external appearance of the component. A rough detection step, an outer electrode rough detection step of detecting a rough position of an electrode row arranged on the outer periphery in the electrode group arranged in a grid, and a rough position of the electrode row arranged on the outer circumference are also performed. And an outer electrode precision detection step of detecting the individual electrode positions of the electrode array arranged on the outer periphery, and calculating the position regulation information necessary for mounting the component from the individual electrode positions of the electrode array arranged on the outer periphery. Using the position regulation information calculation process, this configuration enables high-speed, high-precision extraction of the position regulation information required for mounting in the same process for components with electrodes arranged in a grid pattern in any pattern. Become Sex high position detection becomes possible.

In Japanese Patent Application Laid-Open No. 7-320062, when an image of the bottom surface of a ball grid array type package is taken and image data is supplied, an area not less than a predetermined minimum value and not more than a predetermined maximum value is obtained. The present invention employs a configuration including detection means for detecting a plurality of small-grained images from the image data, and calculation means for obtaining the inclination of the package based on the inclination of a straight line connecting these images. According to this configuration, the ball is detected as a plurality of small-grained images having an area equal to or larger than a predetermined minimum value and equal to or smaller than a predetermined maximum value, and the component posture is grasped by the inclination of a straight line connecting these images. be able to.

[0009]

However, in Japanese Unexamined Patent Publication No. 5-296724, a predetermined area is illuminated, and the ball is judged as a high-luminance portion. For this reason, if there are foreign matters or scratches that look like a ball on the component, these are also displayed with high luminance, so that it is impossible to distinguish the balls from the foreign matters and the like. That is, if there is a foreign matter or the like outside the ball arrangement of the component,
Since such a foreign substance is detected as a ball, the center and inclination of the component cannot be correctly recognized, and there is a problem that the component cannot be finally mounted at an accurate position.

In Japanese Patent Laid-Open No. Hei 6-288732, the centers of all the balls of a component are not detected, but the individual positions of electrode rows arranged on the outer periphery are detected, and the position and inclination of the component are calculated. It has recognized. For this reason, there is a problem that the accuracy of the recognition position is lower than in the case where the position of the component is detected using the center coordinates of all the balls.

In Japanese Patent Application Laid-Open No. 7-320062, means for detecting all the balls first and then performing various processes on them is used for recognizing the component position. For this reason, when the shape of the initially recognized ball is abnormal, there is a problem that the position of the abnormal ball on the ball arrangement cannot be indicated.

Accordingly, an object of the present invention is to provide an image processing method capable of recognizing a ball-shaped object of an electronic component such as a BGA component at high speed and with high accuracy.

[0013]

According to the present invention, there is provided an image processing method for obtaining and processing images of ball-shaped objects arranged on one side of a component vertically and horizontally. The vertical straight line and the horizontal straight line passing through each ball-shaped object are obtained from the image data of the ball-shaped object, and the position of the ball-shaped object obtained from the arrangement information of the ball-shaped object and the vertical straight line and the horizontal straight line are obtained. An inspection area surrounding the image of the ball-shaped object is set in a region including the intersections, and the image of the ball-shaped object in the set inspection area is processed.

According to this configuration, even if a ball-like foreign matter is present at a position other than the ball arrangement, the inspection area is not set at that portion, so that noise can be effectively cut, and Highly accurate image recognition becomes possible.

[0015]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS Hereinafter, the present invention will be described in detail based on embodiments shown in the drawings.

FIG. 1 shows the image recognition method of the present invention in the form of a flowchart. In the first step shown in step S1, the shape and the like of the BGA component in a normal state are taught to the image processing apparatus. In addition to BGA component information such as the normal shape and size of the BGA component to be mounted, the ball pitch of the ball formed at a predetermined position on the base of the BGA component,
All information such as ball shape and ball arrangement information is taught to the image forming apparatus. These pieces of information are reference information for accurately mounting the BGA component.

In the second step of step S2, image data of the BGA component is fetched. FIG. 2 schematically shows an image capturing device for this purpose. On the base 2 of the BGA component 1, a plurality of balls 3 are fixed in a row and a row at predetermined intervals in an array, respectively.
Is picked up by the suction nozzle 4, illuminated from below by the illumination light source 6, captured by the CCD camera (recognition camera) 5, and image-recognized.

That is, in the second step, the BGA component 1 is sucked by the suction nozzle 4 of the component mounting device, and then conveyed to a position above the imaging lens of the CCD camera 5.
The CCD camera 5 has a BGA illuminated by the illumination light source 6.
The base 2 of the component 1 and the balls 3 arranged vertically and horizontally on the base are imaged. At this time, an image such as dust and scratches attached to the base 2 is also displayed as a high-brightness image. Therefore, the image obtained by imaging includes not only the ball 3 but also images such as dust and scratches attached to the BGA component 1. included. For example, FIG. 3 shows an image of the BGA component 1 taken, and the ball 3 is a small-grained image 3a arranged like a white circle vertically and horizontally. In the upper right portion of the figure, an image 3b of a foreign matter that looks like a ball other than the ball 3 is also captured. The image obtained in the second step is temporarily stored as image data.

In the subsequent step S3 of the third step, a straight line passing through the outer row of balls is obtained. In the image data captured in the second step, a range 7 is set for projecting a ball image formed on the BGA component. As shown in FIG. 3, this range is set to an appropriate size surrounding the entire image 1a of the BGA component based on the outer shape of the BGA component.

Then, the projection is performed in the vertical direction indicated by an arrow from the lower side to the upper side of the image 1a of the BGA part shown in the figure. This projection corresponds to the sum of the image data of the captured ball image 3a in the above direction, and a projection waveform 10 as shown in the figure is obtained according to the ball image 3a. Subsequently, the projection waveform 10 is scanned by the detection line 11 from top to bottom. When the detection width W obtained by this scanning comes to correspond to the distance of the outer peripheral ball row obtained from the ball arrangement and the ball pitch taught in the first step, the projection waveforms 10a and 10b at both ends are shifted to the outer peripheral. Since the projection corresponds to the projection of the image of the ball row, the projection range 7
By obtaining the points a and f, the straight lines 12 and 13 that pass almost through the outer row of balls can be obtained.

Subsequently, in step S4 of the fourth step, straight lines passing through all the ball rows are obtained. First, as shown in FIG. 4, a straight line 1 substantially passing through the outer row of balls obtained in the third step
Areas 8a and 8b sufficiently surrounding the outer peripheral portion of the BGA component are set based on 2 and 13. The reason why the area is divided into two as described above is to enable more accurate recognition when the captured image of the BGA component is inclined. Similarly, projection is performed on each of the areas 8a and 8b in the vertical direction, and projection waveforms 14 and 15 are obtained. Subsequently, each projection waveform is scanned from above, and based on the arrangement information of the balls 3 and the information of the pitch of the balls 3 taught in the first step, when the predetermined threshold values L and L 'are reached, this threshold is reached. Values and the projected ball positions a to f on the upper side from the point where the peak waves of the projection waveforms 14 and 15 cross, and the projected ball positions a ′ on the lower side
To f ′ can be obtained respectively.

Similarly, as shown in FIG.
And 9b are set so that the projections are taken in the horizontal direction, and the projected ball positions A to
F and the projected ball positions A ′ to F ′ on the right side are obtained. Then, as shown in FIG. 6, by connecting the positions of the projected balls, straight lines M1 to M1 substantially passing through the ball row in the vertical direction are connected.
M6 and straight lines N1 to N that almost pass through the horizontal row of balls
Ask for 6. In this case, the straight lines M1 to M6 are x = ci * y +
di, and the straight lines N1 to N6 can be represented by y = aj * x + bj (where i, j = 1 to 6, ci, di, a
j and bj are constants).

Subsequently, in a fifth step of step S5, an area 20 for inspecting the ball 3 is set. The inspection area 20 is obtained from the intersection of the straight line passing through the vertical (vertical) ball row and the straight line passing through the left and right (horizontal) ball row obtained in the fourth step and the arrangement information of the balls 3 taught in the first step. The logical product with the position of the determined ball 3 is determined. That is, each ball 3 substantially exists at the position of the intersection of the two straight lines. However, since the ball 3 does not always exist at this intersection, judgment is made only from the intersection of the upper and lower ball rows and the left and right ball rows. Then, it is recognized that the ball 3 exists at the intersection where the ball 3 does not exist. Therefore, by using the arrangement information of the balls 3 taught in the first step together, the inspection area 20 can be set only at the position where the balls 3 actually exist. In this manner, a sufficiently large inspection area 20 surrounding the ball images 3a can be set for the positions of the images of all the balls 3. In this case, since the inspection area 20 is not set in the image 3b such as a foreign substance,
A foreign substance or the like is not erroneously recognized as the ball 3, the recognition time can be shortened, and the position of the BGA component can be recognized with high accuracy.

In the subsequent step S6 of the sixth step, as shown in FIG. 7, each inspection area 2 surrounding each ball image 3a is
For 0, the ball center 3c is determined. The center of the ball is determined by obtaining a pixel corresponding to the ball area from the ball diameter taught in the first step, obtaining a pixel whose brightness is equal to or greater than a predetermined value, and determining a pixel having a brightness equal to or greater than the predetermined value. It can be calculated by calculating the value barycentric coordinates. When the ball center 3c is determined, as shown in FIG. 7, the radius of the ball can be obtained by drawing a plurality of search lines 24 passing through the center 3c, and the shape (deformation) of the ball 3 can be checked. Can be. When a high-luminance image is not captured in the set inspection area 20, it can be determined that the ball 3 is missing.
Furthermore, when the surface of the ball 3 is oxidized, the imaged ball 3 is projected darker than the calculated brightness, so that it is also possible to determine whether the surface of the ball 3 is abnormal.

In the following step S7 of the seventh step, the center of the suction component and the inclination of the suction attitude are obtained. FIG.
Shows an example of calculating the center of one horizontal row of balls, where each circle is the center of the ball obtained in the sixth step, the numbers 1 to 6 are arguments of the array, and there are six ball rows. It is shown that. Since the row of balls extends in the horizontal direction, the Y coordinate is constant, as shown in FIG. 8B, while the X coordinate has a large coordinate value at the center of the ball for each ball row. The regression line 26, X = aM + b, with the argument M of the ball train as an independent variable and the x-axis coordinate of the ball center as a dependent variable
(A and b are constants). Since the center of the argument of the ball array is (1 + 6) /2=3.5, the X coordinate Xc of the center 27 of the ball array is Xc = a × 3.5 +
b. Similarly, the Y coordinate is Yc = a ′ × 3.5
+ B '(for a horizontal row of balls, a' =
0).

Similarly, the center coordinates of the ball rows extending in the vertical direction are similarly obtained, and the centers of the ball rows in the horizontal and vertical directions are plotted with “+” signs, respectively, as shown in FIG. By connecting the points of this "+" mark,
A regression line 28 passing through the center of each vertical row of balls,
Y = A * X + B, and a regression line 29 passing through the center of each ball row in the horizontal direction, X = C * Y + D, can be obtained, and the intersection of the two lines 28, 29 is expressed as Ximg = (BC + D) / (1 −AC), Yimg = (AD + B) / (1−AC), so that the center of the component can be obtained. The component center obtained in this way is shown by an “X” mark in FIG. The suction inclination of the component is
By taking the average of the slopes of the two regression lines 28 and 29,

[0027]

(Equation 1)

Can be obtained as

In the seventh step, the center position of each ball is obtained from the coordinates of the binary centroid, but it is needless to say that it can be obtained by obtaining the second moment as is well known.

[0030]

As is apparent from the above description, according to the present invention, in an image processing method for acquiring and processing images of balls arranged on one surface of a BGA component vertically and horizontally, an image of the acquired balls is obtained. Inspection of obtaining a vertical straight line and a horizontal straight line passing through each ball from the data, and enclosing the ball image in an area including an intersection of the ball position obtained from the ball arrangement information and the vertical straight line and the horizontal straight line. Since the area is set and the image of the ball in the set inspection area is processed, even if a ball-like foreign matter exists at a position other than the ball arrangement, the inspection area is included in that part. Is not set, so that noise can be effectively cut and high-precision image recognition becomes possible.

Further, since the center of the ball array is obtained from the regression line of the ball array, and the center and the inclination of the BGA component are obtained from the intersection and the inclination of the two regression lines at the center of the ball array, an asymmetrical ball arrangement is used. Even if there is, the center and inclination of the BGA component can be obtained.

Further, since the center and the inclination of the BGA component are obtained using all the ball centers, the accuracy of the recognition result is improved as compared with the case where the center and the inclination are obtained using only the outer ball centers.

[Brief description of the drawings]

FIG. 1 is a flowchart showing a flow of an image processing method according to the present invention.

FIG. 2 is a side view showing a schematic configuration of the image capturing device.

FIG. 3 is an explanatory diagram showing a method of acquiring a straight line passing through an outer row of balls.

FIG. 4 is an explanatory diagram showing a method of acquiring a straight line passing through each ball in a vertical direction.

FIG. 5 is an explanatory diagram showing a method for acquiring a straight line passing through each ball in the vertical and horizontal directions.

FIG. 6 is an explanatory diagram showing a method of setting a ball inspection area.

FIG. 7 is an explanatory diagram showing a method of calculating the center of a ball.

FIG. 8 is an explanatory diagram showing how to determine the center of a ball row.

FIG. 9 is an explanatory diagram showing a method of calculating the center and inclination of a BGA component.

FIG. 10 is an explanatory diagram showing a conventional QFP recognition method.

FIG. 11 is an explanatory diagram showing a state in which a conventional BFP component recognition method uses a QFP recognition method.

[Explanation of symbols]

 1 BGA parts 3 Ball 4 Suction nozzle 5 CCD camera 6 Illumination light source

Claims (4)

[Claims] 1. An image processing method for acquiring and processing an image of a ball-shaped object arranged vertically and horizontally on one surface of a component, wherein a vertical direction passing through each ball-shaped object from the acquired image data of the ball-shaped object A straight line and a horizontal straight line are acquired, and an inspection area surrounding the image of the ball-shaped object is set in an area including an intersection of the position of the ball-shaped object obtained from the arrangement information of the ball-shaped object and the vertical straight line and the horizontal straight line. And an image processing method for processing an image of the ball-shaped object in the set inspection area. 2. The method according to claim 1, wherein an image of the ball-shaped object in the inspection area is processed to determine a center position of each ball-shaped object, or the shape of the ball is determined.
The image processing method according to 1. 3. The method according to claim 1, wherein a regression line passing through the center of each ball-shaped object arranged in the vertical or horizontal direction is obtained to obtain the center of the ball array on the straight line. Image processing method. 4. The method according to claim 1, wherein the center of the component image is obtained from the intersection of the regression lines passing through the centers of the ball arrays on the vertical and horizontal lines, and the inclination of the component image is obtained from the inclination of the regression line. The image processing method according to claim 3.