JP3369114B2 - Automatic focusing method and apparatus by image measurement - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an automatic focusing method and apparatus which can be suitably used when performing non-contact strain measurement and the like using image processing. INDUSTRIAL APPLICABILITY The present invention can be suitably used, for example, when detecting the distortion of the lead portion of a semiconductor package, for example, a quad flat package (QFP).

The non-contact type strain measurement technique using image processing is, for example, a technique for obtaining the strain amount by measuring the amount of strain before and after applying the strain weight on the printed board with an image. Is. For this distortion measurement, an automatic focusing function is essential to obtain a clear image. In addition, after the distortion is weighted, it is necessary to detect the position corresponding to the measurement point before the distortion is weighted, in addition to the automatic focusing.

[0003]

2. Description of the Related Art As prior arts related to the present invention, Japanese Patent Laid-Open Nos. 6-14243 and 7-87378 describe a focusing based on a contrast ratio. In Japanese Patent Laid-Open No. 6-14243, a scanning direction in which suitable contrast information is obtained is detected, and information relating to a high frequency component in video information obtained by scanning in the detected direction is used as a focus evaluation value. It is disclosed.
Further, in Japanese Patent Application Laid-Open No. 7-87378, a pattern in a direction inclined with respect to the optical axis is projected on a target object, and a signal corresponding to the projected pattern image is detected in detecting the image of the target object. The focus position is detected based on the position where the contrast is maximum.

However, these prior arts can deal only with the case where the object is a plane and the focal position is one. That is, focus detection is effective only for the entire image. In Japanese Patent Laid-Open No. 8-201022, a captured image is subjected to a first-order differential filter, the image is binarized, the binarized image is thinned, and the total number of all pixels on the screen is characterized. I ask for it as a quantity. However, in this case, since the result of the first-order differentiation method is used as it is for the focus result, it is not possible to limit a plurality of focus points or focus areas.

Further, Japanese Laid-Open Patent Publication No. 6-133203 discloses that horizontal scanning lines of the image pickup screen within the focus area are selected or scanning lines are thinned out.

[0006]

In the above-mentioned prior art,
For example, before and after applying distortion weight to the object,
It was not possible to detect characteristic points and focus on a plurality of detected focus positions with respect to an object having a plurality of focus positions. There are the following problems in automatic focusing before and after distortion weighting of an object.

That is, in the automatic focusing before applying the distortion weight, it is not possible to specify a peculiar point in advance, and therefore it is necessary to automatically focus the characteristic portion of the image.

[0008]

In order to achieve the above object, according to the present invention, an image pickup means for picking up an image of an object to be measured and a variable distance for changing the distance between the image pickup means and the object to be measured. And a control unit for controlling the distance varying unit, which has a function of inputting an image captured by the image capturing unit, and applies a distortion to the measurement target.
In the state before overlapping, the input image at each distance position within the range of the arbitrary distance is scanned in the focus area of the specific range, and the value obtained by adding the edge strength value in the focus area is calculated. Obtained as the focus degree, compare the focus degree at each scanning location, detect the location of the maximum focus value, compare the detection location at each of those distance positions, and determine the location of the maximum focus value. The position that includes is the focus position, and the detected position at the focus position
Images can be stored as a matching dictionary file.
And in the state after the strain is applied to the measurement target,
Use the matching dictionary file for each distance position
Matching with the input image of the
An automatic focusing method by image measurement is provided, which is characterized by measuring a position and a matching degree .

As the edge strength value, a value obtained by applying a Sobel filter is used. Further, it is characterized in that the average value of the edge strength values in the focus area is obtained as the focus degree. Further, it is characterized in that the focus area is fixed to the detected position of the focus area to perform more detailed automatic focusing on the distance position. Further, it is characterized in that the scanning time of the focus area is thinned to shorten the output time. Furthermore, after storing the vertical, horizontal, and height positions of the detected focus coordinates and the image pattern in the dictionary file and applying distortion to the measurement target,
The feature is that auto-focusing is performed by matching based on the dictionary file.

Further, according to the present invention, an image pickup means for picking up an image of a measuring object, a distance varying means for changing a distance between the image pickup means and the measuring object, and a function for inputting an image picked up by the image pickup means. And a control means for controlling the distance varying means and an input image for each distance position within a range of an arbitrary distance to scan a focus area of a specific range, and within the focus area, an edge strength value. The value obtained by adding is calculated as the degree of focus, the degree of focus at each scanning location is compared, the location with the maximum focus value is detected, and the location detected at each of these distance positions is compared to determine the maximum. Measuring means for setting a focus position at a position including a focus value detection position of the
In the state before applying the distortion to the object, the focus position
The image of the detected location in
Storage means for storing the
And the measurement target after distortion is weighted,
Use the matching dictionary file for each distance position
Matching with the input image of the
There is provided an automatic focusing device by image measurement, comprising: a measuring unit that measures a position and a matching degree .

[0011]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described in detail below with reference to the accompanying drawings. FIG. 1 shows an image measuring device to which the present invention is applied.
It is used when automatic measurement of total bin distortion of FP is performed.
In the figure, 1 is a test printed circuit board to be measured,
Reference numeral 2 is a dedicated jig for fixing the measurement target, 3 is an XY stage capable of moving the jig in the horizontal direction (XY direction) to position the measurement target 1, and 4 is the X-axis.
It is a stage control device for controlling the Y stage. Reference numeral 5 is a microscope, 6 is an objective lens of the microscope, 7 is a light source, 8 is a CCD camera, and 9 is a Z stage that can move these in the vertical direction with respect to the measurement target 1.
The stage is connected to the Z stage control device 10, and the light source 7
Is connected to the light amount adjusting device 11, and the CCD camera 8 is connected to the image input / image processing board 12 of the control device 20. Stage control devices 4 and 10 and light amount control device 11
Is also connected to the control device 20.

FIG. 2 shows a state in which the test printed circuit board 1 to be measured is placed on the dedicated jig 2 before straining and after straining. As shown in the drawing, when the strain-weighted shaft 13 at the lower center of the special jig 2 is moved upward, the special jig 2 is fixed at both upper ends of the fixing shafts 14a and 14a.
The printed board 1 fixed to the dedicated jig 2 is also deformed with respect to b. It should be noted that FIG. 2 shows the distortion deformation in an exaggerated manner than it actually is.

In the present invention, two types of automatic focusing methods are executed before and after distortion weighting of the object to be measured. In automatic focusing before distortion weighting, specific points cannot be specified in advance, so it is necessary to automatically focus on the characteristic parts of the image, and automatic focusing after distortion weighting (matching) You must also consider that.

First, automatic focusing before distortion weighting will be described. In this regard, the principle underlying the present invention will be described. First, as a simple example, a clear black and white image as shown in FIG. 3A will be described as an example. When the image is in focus, the P1-P2 profile of this image is shown in FIG.
As shown in FIG. 3B, the edge of the boundary portion rises sharply, and when it is out of focus, the rising edge also becomes gentle as shown in FIG. 3C. The steepness of the edge is obtained by using a differential filter in the original image (a Sobel filter is shown in FIG. 5 as an example).
It is obtained as an edge strength value by multiplying by. The principle of automatic focusing of the present invention is to judge the edge strength value to find the in-focus position of the image.

In automatic focusing, an image is input while the height (Z) stage 9 is moved in steps. A value obtained by summing the edge strength values of the entire input image or an average value is obtained, and this is used as the focus degree. The focus degree thus obtained is compared with the images obtained at each height, and the height at which the image having the maximum focus degree is measured is set as the focus position. At this time, the processing amount is too large for the entire input image,
Since the range is large, the characteristic part is blurred. Therefore, the window of the specific range is usually fixed at a fixed position, and the process is advanced in that area.

FIG. 5 shows the case where a Sobel filter is used as an example of the differential filter. The Sobel filter is, for example, E = | C + 2F + 1-A-2D-G | + | G + 2H + 1 for an image as shown in the upper left of FIG.
A matrix of -A-2B-C | is applied. As a result, the degrees of gray-scale angles in the vertical direction and the horizontal direction are added to emphasize the edge.

By the way, the actual measurement image is shown in FIG.
It is not so simple as described above, and there are a plurality of edge components as shown in the micrograph of FIG. 4, for example. Further, when the image of the lead portion of the QFP mounted on the printed board is imaged, the surface is uneven due to the degree of application of the solder, and the object is not a flat surface, so there are multiple focusing points. Therefore, in the above-mentioned “window area fixing method”, when there is unevenness in solder application, the surface may not be in focus, or an inappropriate portion may be in focus.

Further, in the measurement before distortion weighting, it is necessary to create a matching dictionary for measurement after distortion weighting and to specify the measurement location of the matching dictionary. In order to solve these problems, the present invention has devised the following autofocus (automatic focusing) method. FIG. 6 shows the input image and FIG.
Is a flow chart of the measurement logic before distortion weighting. The input image 21 is, for example, a CCD camera 8 having 640 × 480 pixels.
The measurement area 22 is an image area input to (FIG. 1).
Is an area for measurement of, for example, 60 × 60 pixels. Figure 6
As shown in (b), the entire input image 21 is scanned in this focus area 22. For the focus degree, the Sobel value for each pixel in each measurement focus area 22 is added, and the value or the focus area 22
It is a value divided by the number of pixels of.

The position where the degree of focus obtained by scanning the measurement area 22 in the entire input image 21 is the maximum is the focus area 22a of the maximum degree of focus of the input image 21. At the same time, the focus degree, the position of the focus area 22a, and the height position are stored in the memory. At each height, this is done and the focus degrees in memory are compared and the maximum height is taken as the focus position.

Next, the measurement logic before distortion weighting will be described with reference to the flowchart of FIG. First, the height (Z)
The stage 9 (FIG. 1) is moved to the starting point height of the scanning range.
Next, input an image and input an image (for example, 640 × 48).
0 pixel) is stored in the image memory 1. Next, for the input image, an edge enhancement filter (eg Sobel filter)
The value multiplied by is stored in the image memory 2. A focus area (for example, m × n = 60 × 60 pixels) is set at the upper left of the screen of the input image. The total value or the average value of the values applied with the edge enhancement filter (for example, Sobel filter) in the focus area is calculated as the focus degree. If the calculated detected focus degree is greater than the maximum focus degree,
The detected focus degree at this time is reset as the maximum focus degree. Unless the position of the measurement area is at the end of the image, the measurement area is laterally moved (scanned) by one pixel, and the focus degree (f = Σe (x,
y) / m × n) is calculated, and this operation is repeated until the end of the image, that is, over the entire image.

In the same input image, the position coordinates of the focus area where the degree of focus is maximum, the height position of the height (Z) stage, and the degree of focus are stored in the AF determination memory. Unless the position of the height (Z) stage is the end point height of the scanning range, the height (Z) stage is moved by a predetermined step distance (for example, 0.2 μm) and the same operation is repeated.

Next, the focus degrees at the respective heights stored in the AF determination memory are compared, and the maximum focus degree and the corresponding focus area position and height (Z) stage position are extracted. Next, the height (Z) stage is moved to the detected height position. An image is input here, and the input image (for example, 640 × 480 pixels) is stored in the image memory 1.
A window based on the detected position of the focus area is cut out from the input image, and the image pattern in the cut out range is stored in a file as a matching dictionary. Then, the position of the height (Z) stage, the position of the XY stage, and the position of the focus area are stored in the measurement file.

8 and 9 show the measurement logic before distortion weighting.
FIG. Microscope 5
Input image 21 captured by the CCD camera 8 by (FIG. 1)
Is the height h of the height (Z) station 9.₁, H₂,, change
Are measured sequentially. For each input image 21,
While moving (scanning) the rear 22 (m × n pixels),
Measure the degree of focus. Here, the height h₃In position
Maximum focus degree (fh _3-max= 100) is obtained.

A feature of the present invention is to find a portion having a high Sobel value in each input image as described above. As a result, it is possible to perform appropriate focusing on an object whose surface is uneven. As shown in the examples of FIGS. 8 and 9, there are two measurement targets at the focus positions, and the focus areas are A and B, respectively. In the case of autofocus in which the focus area is fixed, if the focus area is first set in the area A, the focus cannot be focused in the area B having a higher degree of focus. FIG. 10 is a graph showing the transition of the focus degree in this case.
It becomes like (a). In the present invention, in each input image, since the location with the highest degree of focus will be searched,
Area B can also be detected. FIG. 10B is an example of a graph of focus degree transition according to the present invention.

[0025] That is, in FIG. 10 (a), the height (Z) to be changed the position of the stage is measured area A is fixed, the maximum degree of focus at the position of height h ₂ (fh _2- It is determined that _max = 70) appears. On the other hand, in the present invention, since the measurement area is scanned in each input image in which the position of the height (Z) stage is changed, the maximum focus degree (fh _3-max = 100 in area B of height h ₃ ). ) Can be determined to be obtained.

FIG. 11 is a flowchart of the measurement logic after distortion weighting. First, the principle of the present invention will be described. The image of the location detected by autofocus before distortion weighting is stored in the file as a matching dictionary file. At this time, the characteristic point temporarily stored in the dictionary is set as point A. After distortion weighting, matching is performed at each step height based on this dictionary, and the matching position and matching degree are measured. Among them, the position with the highest matching degree is focused, and the height position thereof becomes the focus position, and the position in the Z direction of point A and the XY matching position become the XY position of point A. The amount of deformation of the measurement target due to the strain is calculated by comparing the two measurement data before and after the strain weighting.

That is, in FIG. 11, after loading the matching dictionary file, the height (Z) stage is moved to the starting point height of the scanning range. Next, an image is input, and the input image (for example, 640 × 480 pixels) is stored in the image memory 1. Next, the input image is matched with the matching dictionary. In matching, there is a matched area, and the matching degree and height (Z) stage position, when the matching degree becomes the maximum in the scanning so far,
The focus area position is stored and repeated until the height (Z) stage reaches the end position height position of the scanning range.

The matching degree when the matching degree is maximum,
And the height (Z) stage position and focus area position are pulled out. Height detected (Z) at the stage position
Move the stage. At this position, the matching degree and the height (Z) stage position, the XY stage position, and the focus area position are stored in the measurement file.
The embodiments of the present invention have been described above in detail with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments, and various forms, variations, and modifications within the spirit and scope of the present invention It should be noted that etc. are possible.

[0029]

As described above, according to the present invention, the image pickup means for picking up the image of the measuring object, the distance varying means for changing the distance between the image pickup means and the measuring object, and the image pickup means. And a control means for controlling the distance varying means, which has a function of inputting an image picked up in, and an input image for each distance position within a range of an arbitrary distance, and a focus area of a specific range is set. The value obtained by scanning and adding the edge strength value in the focus area is obtained as the degree of focus, the degree of focus at each scanning location is compared, the maximum location is detected, and detection at each of these distance positions is performed. Since the points are compared and the maximum position is set as the focal position, the present invention can be applied even when the surface of the measurement target has irregularities and there are a plurality of focal positions. Further, in the present invention, in the case where a strain weight is given to the measurement target,
In the measurement after distortion weighting, autofocus can be performed using a dictionary (matching dictionary) before distortion weighting.

[Brief description of drawings]

FIG. 1 shows an image measuring device when the present invention is applied.

FIG. 2 shows a test printed circuit board to be measured, which is set in a dedicated jig, before and after strain-weighting, respectively.

FIG. 3 shows a case where the edge of the boundary portion is steep (b) and a case where the edge is gentle (c) with respect to the image profile (a).

FIG. 4 is a photograph of a surface of a measurement target having irregularities.

FIG. 5 is a diagram for explaining a Sobel filter.

FIG. 6 shows an input image and a focus area (a),
Further, it shows a state in which the input image is scanned with the focus area.

FIG. 7 is a flow chart of measurement logic before weighting.

FIG. 8 schematically shows a state of autofocus of the measurement logic before weighting.

FIG. 9 schematically shows a state of autofocus of the measurement logic before weighting.

FIG. 10 shows a case where the focus area is fixed (a),
The case (b) of scanning the focus area as in the present invention is shown for comparison.

FIG. 11 is a flow chart of measurement logic after weighting.

[Explanation of symbols]

1 ... Measurement target 2 ... jig 3 ... XY stage 4, 10 ... Stage control device 5 ... Microscope 6 ... Objective lens 7 ... Light source 8 ... CCD camera 9 ... Height (Z) stage 11 ... Light intensity adjusting device 12 ... Image input / output port 13 ... Strain-weighted shaft 20 ... Control device

─────────────────────────────────────────────────── ─── Continuation of the front page (51) Int.Cl. ⁷ Identification code FI G02B 21/26 G02B 7/11 DJ (72) Inventor Tetsuo Hizuka 4-1-1 Kamiodanaka, Nakahara-ku, Kawasaki-shi, Kanagawa Fujitsu (56) References JP-A-4-150176 (JP, A) JP-A-5-19161 (JP, A) JP-A-9- 211307 (JP, A) (58) Fields investigated (Int. Cl. ⁷ , DB name) G02B ^7/ 28-7/40

Claims (4)

(57) [Claims] 1. An image pickup means for picking up an image of a measurement target, a distance changing means for changing a distance between the image pickup means and the measurement target, and a function of inputting an image picked up by the image pickup means, A control means for controlling the distance varying means, before applying a strain to the measurement object.
In this state, the input image at each distance position within the range of the arbitrary distance is scanned in the focus area of the specific range, and the value obtained by adding the edge strength value in the focus area is obtained as the focus degree. , Comparing the degree of focus at each scanning location, detecting the location of the maximum focus value, comparing the detection location at each of those distance positions,
The focus position is defined as the position that includes the detected location of the maximum focus value, and the image of the detected location at the focus location is marked.
Storing as a dictionary file for
After applying distortion to the target,
Use the dictionary file for
Matches with the input image and matches with the matching position
An automatic focusing method based on image measurement, which is characterized by measuring the tangency. 2. The automatic focusing method by image measurement according to claim 1, wherein a Sobel filter is applied and a liquid value is used as the edge strength value. 3. The automatic focusing method according to claim 1, wherein an average value of edge strength values in the focus area is obtained as a focus degree. 4. An image pickup means for picking up an image of a measuring object, a distance varying means for changing a distance between the image pickup means and the measuring object, and a function of inputting an image picked up by the image pickup means, With the control means for controlling the distance varying means and the input image for each distance position within the range of the arbitrary distance, the focus area of the specific range is scanned, and the edge strength value is added within the focus area. Value is calculated as the degree of focus, the degree of focus at each scanning location is compared, the location with the maximum focus value is detected, and the locations detected at each distance position are compared to determine the maximum focus value. An autofocus executing means for setting a focus position at a position including a detection point, and distortion for the measurement target.
In the state before weighting,
Store the image of the selected part as a dictionary file for matching
Storing means and means for applying strain to the measurement target
And the matching for the measurement target after distortion is weighted
For each distance position, use the dictionary file for
Matches with force image and matches with matching position
An automatic focusing device by image measurement, comprising: a measuring unit that measures a degree of blur.