JPH08189822A - Inclination detecting method - Google Patents

Abstract

PURPOSE: To specify the inclination of an object based on the distribution of a table by setting the table of plurality of cells composed of intercept values of line components extended to a projection line and inclination from picture elements in rectilinear components extracted from a raw image and adding the values of the relevant cells while searching the picture elements. CONSTITUTION: A chip 3 is photographed by a camera 10 and a raw image after processed by an A/D converter 11 is stored in a raw image memory 13. A rectilinear data extracting part 15 extracts rectilinear components in nearly transverse direction from the raw image data and stores the results in a processed image memory 14. A θY table 17 is composed of a plurality of cells having as components; the inclination θ and an intercept value Y of a line component extended to a constant projection line from a picture element on a straight line. A table processing part 16 carries out addition for relevant cell values of the table 17 one by one while searching the picture elements on a straight line. An inclination judging part 18 computes an approximate curved line by the method of least squares according to the table 17 for which processing by the processing part 16 has been completed and computes the inclination of an object from an inflection point.

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 inclination of an object observed by a camera, such as electronic parts and boards.

[0002]

2. Description of the Related Art In recent years, it has been widely practiced to observe an object with a camera and detect a displacement of the object from an ideal position. This positional deviation includes positional deviation in the θ direction in the horizontal plane (referred to as “tilt” in this specification).

A conventional inclination detecting method will be described below with reference to FIGS. FIG. 11 is a plan view illustrating a conventional object. In FIG. 11, 1 is an adhesive sheet 2
3 is a chip as an object stuck on the upper surface of the adhesive sheet 2. The conventional tilt detection method is an application of the pattern matching method, and as shown in FIG. 11, a window 4 for setting a reference pattern at an ideal position without tilt is set so as to surround the chip 3. .

FIG. 12 is an enlarged view of the window shown in FIG. As shown in FIG. 12, in order to detect the inclination of the chip 3, it is sufficient to specify the positions of the edges C1, C2, C3, C4 of the corners 3a, 3b, 3c, 3d of the chip 3.
Therefore, in this example, the edges C1, C2, C3, C
Reference patterns 5, 6, 7, and 8 having P4 as the center P1, P2, P3, and P4 are set. Then, when the image of the actual object is obtained, a position where these reference patterns 5, 6, 7, and 8 best match is found, and the centers C1, C2, C3, and C4 at the positions are found at the edge C1 of the chip 3. , C2, C3, C4.

[0005]

FIG. 13 is a plan view illustrating a conventional object. FIG. 13 shows a case where there is a tilt, not the ideal position (no tilt) as shown in FIG. Then, FIG. 14 is a partially enlarged view of FIG. 13, and shows how the reference pattern 5 is interpreted as matching the corner portion 3a. Then, when the matching between the reference pattern 5 and the image of the corner 3a is maximized,
As shown in FIG. 14, the center C of the reference pattern 5
1 (which is the position of the processing edge P1) deviates from the actual edge P1. Therefore, the inclination of the chip 3 calculated based on the center C1 does not match the actual inclination of the chip 3. As described above, the conventional inclination detection method has a problem that the inclination detection accuracy is low.

Therefore, an object of the present invention is to provide an inclination detecting method capable of detecting the inclination of an object with high accuracy.

[0007]

A tilt detecting method according to the present invention comprises a step of obtaining a raw image of an object, a step of obtaining a processed image in which a straight line component is extracted from the obtained raw image, and pixels on a straight line. Set a table consisting of multiple cells that include the intercept value of the line segment and the slope of the line segment, which are extended to a certain projection line from, and search the pixels on the straight line to find the value of the corresponding cell among the cells. It has a step of summing products and a step of identifying the inclination of the object based on the distribution of the table.

[0008]

With the above configuration, a processed image in which a straight line component is extracted from a raw image is obtained, and a plurality of lines each having an intercept value of a line segment extending from a pixel on this straight line to a projection line and a slope of the line segment are used as elements. A table consisting of cells is set. Then, the pixels on the straight line are searched, the values of the corresponding cells are summed, and the inclination of the object is specified by the distribution of the values of the cells. Here, when the object has an inclination, the overall tendency of the object is directly reflected in the distribution of the cell values. Therefore, the detection accuracy can be remarkably improved as compared with the conventional inclination detection method in which the inclination is detected only by local matching.

[0009]

Embodiments of the present invention will now be described with reference to the drawings.

FIG. 1 is a block diagram of a visual inspection apparatus according to an embodiment of the present invention. The same components as those shown in FIG. 11 and subsequent drawings showing the conventional means are designated by the same reference numerals and the description thereof will be omitted.

In FIG. 1, 10 is a camera for observing the chip 3 as an object, 11 is an A / D converter for digitizing an image of the camera 10, and 12 is a control unit for controlling the whole. Further, 13 is a raw image memory for storing a raw image output from the A / D converter 11, and 14 is a processed image obtained by performing a Sobel transformation on the raw image by the straight line data extraction unit 15 and extracting only a straight line component. The processed image memory 16 processes the θY table 17 composed of a plurality of cells whose elements are the intercept value (Y) of the line segment and the inclination (θ) of the line segment extending from the pixels on the straight line to a certain projection line. The table processing unit,
18 is an approximate curve obtained by the method of least squares based on the θY table 17 after the processing of the table processing unit 16 is completed,
It is an inclination determination unit that obtains the inclination of the object from the inflection point of this approximate curve.

Next, with reference to FIG. 2 (a flowchart showing a tilt detecting method according to an embodiment of the present invention), an overall flow of the tilt detecting method will be described. First, in step 1, the camera 10 takes an image of the chip 3 and the A / D converter 11
The raw image digitized at is stored in the raw image memory 13. Next, in step 2, the straight line data extraction unit 15 performs Sobel transformation on the raw image in the raw image memory 13,
Only the substantially straight line components are extracted. That is, in this example, only the edge portion of the chip 3, the wiring, and the like that are substantially parallel to the X axis are extracted.

FIGS. 3A to 3C are explanatory views of the straight line extraction processing of the inclination detecting method in the embodiment of the present invention. Here, when a raw image having an inclination with respect to the XY axes is obtained as shown in FIG. 3A, the straight line data extraction unit 13
By using a 3 × 3 spatial filter as shown in FIG. 3B, only substantially straight line components as shown in FIG. 3C are extracted. Here, the Sobel transform itself is a well-known technique in the image processing field, and therefore a detailed description thereof will be omitted.
Here, in this example, the lower left corner point of the window 4 is (0, Y
min) and the upper right corner point is (Xmax, Ymax) as a rectangular area, and (0, Ymin) to (0, Yma)
The left side of the window 4 up to x) is the projection line 19.

Then, the table processing section 16 performs the table processing of step 3 of FIG. 2, and the tilt determination section 18 performs the tilt determination of step 4. Here, the processing content of step 3 will be described in detail later with reference to FIG. 7 and the processing content of step 4 will be described later with reference to FIG.

Before explaining the processing contents, the intercept Y and the inclination θ of the projection line 19 will be described with reference to FIG. 4 (an explanatory view of the intercept and the inclination of the inclination detecting method in one embodiment of the present invention). Here, in this embodiment, as the inclination θ, only the constant step values θ1, θ2, θ3, θ4, and θ5 are used. Since the straight line components extracted as described above are almost parallel to the X axis, the step may be about 1 to 5 °. In addition, θ1>θ2> θ3
= 0>θ4> θ5. Then, as shown in FIG. 4, when there is a pixel Pn (Xn, Yn) on a straight line, the intercept Y _k of the projection line 19 when the inclination θ _k ( _k is a subscript) is taken is Y _k = int (Xn · tan θ _k + Yn).

Next, the relationship between the inclination θ _k and the inclination of the extracted straight line component will be described with reference to FIGS. 5A to 5C (an explanatory view of the inclination in an embodiment of the present invention). First, as described above, the inclination θ _k has a discrete value set by software regardless of the inclination of the object. Therefore, a certain inclination θ _k may be largely deviated from the inclination of the object, and a certain inclination θ _k may be completely coincident with the inclination of the object.

Here, as shown in FIG. 5A, the inclination θ
_{If k} 1 deviates significantly from the slope of the extracted straight line, the line segment with the slope θ _k 1 from the pixel on the straight line (see arrow)
When is extended to the projection line 19, the intercepts from each pixel have different values. Therefore, if the number distribution for each intercept is taken, the pattern becomes gentle and wide.

On the other hand, as shown in FIG. 5C, when the inclination θ _k 3 happens to coincide with the inclination of the extracted straight line,
The line segments extending from each pixel to the projection line 19 overlap, and the intercepts of each pixel on the same straight line completely match. Therefore, the distribution of intercepts in the case of FIG. 3C becomes steep. When there is an intermediate matching relationship as shown in FIG. 3 (b), the distribution of intercepts is also shown in FIGS. 3 (a) and 3 (c).
It will be something in between.

In this embodiment, paying attention to the above relationship,
A θY table 17 shown in FIG. 6 (configuration diagram of the θY table in one embodiment of the present invention) is provided, and processing based on this is performed. The θY table 17 has a tilt θ _{k as} the first element.
( _K = 1, 2, ..., 5), the intercept Y (Y =
Ymin, Ymin + 1, ..., Ymax) 2
The dimension array is Val (,). Each element Val (θ _k , Y) is called a cell, and a cell group having the same inclination θ _k is called a segment.

Next, the operation of the table processing section 16 will be described with reference to FIG. 7 (a flow chart showing the table processing in one embodiment of the present invention). First, in step 30, the value of each cell Val (,) is initialized to 0, and in step 31, the search position Pn is set (initialized) at the upper left point (0, Ymax) of the window 4.

Thereafter, in step 32, this search position P
Until n reaches the end position (Xmax, Ymax), it is examined whether or not it is a black pixel (the search position Pn exists on the pixel on the extracted straight line) (step 33). If it is not a black pixel, the search position Pn (Step 34) is repeated. That is, the pixels on the extracted straight line are searched. If the current search position Pn is a black pixel, the process after step 35 is performed, and the process returns to the search loop after step 32.

Now, if the search position Pn is a black pixel,
Subscript at step 35_k Is initialized to 1 and the subscript_k Is from 5
Large (that is, inclination θ1 to θ_k Process is completed)
Subscript _k Present inclination θ_k About the projection line 19
To Y_k (Step 37), the θY table
Corresponding cell Val (θ_k , Y_k ) Value of 1
Add (step 38), subscript_k Increment
(Step 39) and repeat the processing from step 36 onward.
Return. From the above, the corresponding cell Val (θ_k , Y_k )of
The value is incremented by one.

FIG. 8 is a view showing an example of the θY table after addition in the embodiment of the present invention. When the process of FIG. 7 is completed, the value of each cell has a certain distribution, and the overall characteristic of this distribution indicates the inclination of the chip 3 which is the object. In FIG. 8, the shaded cell value indicates the maximum value in each segment. Next, the operation of the tilt determination unit 18 will be described with reference to FIG. 9 (flow chart showing tilt determination in one embodiment of the present invention) and FIG. 10 (illustration diagram of an approximate curve in one embodiment of the present invention). First, in step 40, the maximum value is extracted from the values of all cells and is set as the maximum value W _M. Then, the gradient of the cell having the maximum value W _M is set as the maximum gradient θ _M (step 41).

Next, the maximum value of the cell in the segment on the left side of the cell having the maximum value W _M is obtained, and this is calculated as the maximum value W on the left side.
_L , and the slope at that time is the left-side maximum value slope θ _M-1 (step 42). Similarly, for the segment on the right side of the cell having the maximum value W _M , the right side maximum value W _R and the right side maximum value slope θ
_{M + 1} is calculated (step 43).

Next, as shown in FIG. 10, the horizontal axis is tilted by θ,
The θ-W plane with the maximum value W on the vertical axis is hypothesized and
In the −W plane, the point (θ _M-1 , W _L ) and the point (θ _M ,
W _M ), the point (θ _{M + 1} , W _R ) is used to obtain an approximate curve S (step 44). Then, the inflection point Q (θ _R , W _R ) is obtained by differentiating the approximate curve S, and the θ coordinate θ _R of the inflection point Q is used as the inclination to be obtained (step 45). As is clear from the above description, the slope θ _k
Was originally set as a discrete value, but even if the correct slope is an intermediate value, the slope can be accurately detected by using the above approximate curve. Further, no matter how the object is tilted, the overall tendency of the tilt is reflected in the value of each cell of the θY table 17, and accurate tilt detection can be performed. it can.

The nozzle is rotated according to the detected inclination to perform correction, and the nozzle is sucked and mounted on the substrate. And
The next chip tilt detection, suction, and mounting are repeated.

[0027]

The inclination detecting method of the present invention comprises the steps of obtaining a raw image of an object, obtaining a processed image in which straight line components are extracted from the obtained raw image, and performing a constant projection from pixels on the straight line. A step of setting a table made up of a plurality of cells including, as elements, the intercept value of the line segment extended to the line and the slope of the line segment, and adding the value of the corresponding cell among the cells while searching for pixels on a straight line; , And the step of specifying the inclination of the object based on the distribution of the table, the accuracy of the inclination detected by the inclination of the object does not decrease, so that accurate inclination detection can be realized. .

[Brief description of drawings]

FIG. 1 is a block diagram of a visual inspection apparatus according to an embodiment of the present invention.

FIG. 2 is a flowchart showing a tilt detection method according to an embodiment of the present invention.

FIG. 3A is an explanatory diagram of a straight line extracting process of a tilt detecting method according to an embodiment of the present invention. FIG. 3B is an explanatory diagram of a straight line extracting process of a tilt detecting method according to an embodiment of the present invention. Explanatory drawing of straight line extraction processing of the inclination detection method in one embodiment

FIG. 4 is an explanatory diagram of an intercept and an inclination of an inclination detection method according to an embodiment of the present invention.

5A is an explanatory diagram of a tilt in one embodiment of the present invention. FIG. 5B is an explanatory diagram of a tilt in one embodiment of the present invention. FIG. 5C is an explanatory view of a tilt in one embodiment of the present invention.

FIG. 6 is a configuration diagram of a θY table according to an embodiment of the present invention.

FIG. 7 is a flowchart showing table processing according to an embodiment of the present invention.

FIG. 8 is an exemplary diagram of a θY table after addition in one embodiment of the present invention.

FIG. 9 is a flowchart showing tilt determination in one embodiment of the present invention.

FIG. 10 is an exemplary diagram of an approximated curve in one embodiment of the present invention.

FIG. 11 is a plan view illustrating a conventional object.

FIG. 12 is an enlarged view of the window shown in FIG.

FIG. 13 is a plan view illustrating a conventional object.

FIG. 14 is a partially enlarged view of FIG.

[Explanation of symbols]

 3 chips 17 θY table 19 projection line Q inflection point

Claims (3)

[Claims] 1. A step of obtaining a raw image of an object, a step of obtaining a processed image in which a straight line component is extracted from the obtained raw image, and a line segment extending from a pixel on the straight line to a certain projection line. A step of setting a table composed of a plurality of cells including an intercept value and the inclination of the line segment as an element, and adding values of corresponding cells among the cells by a constant value while searching for pixels on the straight line; And a step of identifying the inclination of the object based on the distribution of the table. 2. The tilt detection method according to claim 1, wherein the processed image is a Sobel transform of the raw image. 3. The slope is specified by referring to an inflection point of an approximate curve determined from the maximum value of the distribution of the table and the maximum value of each cell group adjacent to the maximum value. The tilt detection method according to claim 1, wherein