JP3111432B2 - Image processing device - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image processing apparatus for positioning a semiconductor device, and more particularly to an improvement in the accuracy of the image processing apparatus.

[0002]

2. Description of the Related Art A manufactured IC is mounted on a mounting object such as a substrate. For that purpose, it is necessary to transport the lead portion of the IC at a specified position of the mounted object in a specified posture.

In general, methods for measuring the position and orientation of an IC from image data read by an imaging device include a method using the center of gravity and a principal axis angle of the entire IC (hereinafter referred to as an overall reference method), and a method for measuring the leading end of the IC. (Hereinafter referred to as the average straight line method) is known.

In the overall reference method, image data read by an image pickup device is binarized to determine the center of gravity of image data of the entire IC. As for the principal axis angle, an approximate ellipse circumscribing the image data of the entire IC is assumed, and the calculation is performed from the major axis and the minor axis of the ellipse. Thus, the center of gravity and the spindle angle of the entire IC are calculated. Position adjustment is performed based on the calculated center of gravity, and angle adjustment is performed based on the main shaft angle. Thereby, the lead portion of the IC can be accurately mounted on the substrate.

The average straight line method binarizes image data read by an imaging device and sequentially scans the data to sequentially detect the positions of the leading ends of the ICs and minimize the average straight line passing through each leading end. Determined by the square method or the like. From the obtained average straight line, the position and orientation of the IC (the same figure, rotation angle θ) are obtained.

[0006]

However, the following problems have been encountered in the above-described overall reference method and average straight line method.

[0007] Generally, the external dimension accuracy of an IC package is not very good. For this reason, errors occur in the center of gravity and the spindle angle obtained by the overall reference method, and the IC cannot be mounted accurately. If the shape of the IC package is close to a square, the major axis and the minor axis of the circumscribed approximate ellipse cannot be distinguished, and an accurate principal axis angle cannot be obtained. Further, measurement errors occur due to image noise and dust present around the IC to be measured.

On the other hand, in the average linear method, when detecting the position of the leading end of the IC, there is a quantization error in binarization, and there is a possibility that the measurement accuracy may vary.
In particular, when the width and pitch of the leads of the IC are miniaturized, there is a problem that the above variation cannot be ignored.

SUMMARY OF THE INVENTION The present invention solves the above-described problems, and increases the position and orientation of a semiconductor device even when the shape of the semiconductor device is close to a square, or when image noise or dust existing around the semiconductor device is present. It is an object of the present invention to provide an image processing device capable of measuring with high accuracy.

[0010]

According to a first aspect of the present invention, there is provided an image processing apparatus, comprising: a lead tip temporary position measuring means for measuring a temporary lead tip position for each lead of a semiconductor device based on binary image information; A true lead tip position calculating means for calculating a true lead tip position for each lead from a grayscale image of the lead tip position and a surrounding area, and a semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. A center-of-gravity calculating means for calculating a center of gravity and an inclination.

According to a second aspect of the present invention, in the image processing apparatus, the semiconductor device is further mounted such that the arrangement direction of the leads substantially coincides with the horizontal scanning direction. In the scanning direction and at the base of the lead,
Move by a predetermined pixel, detect adjacent leads by scanning in the horizontal scanning direction, and then move in the vertical scanning direction and in the leading direction of the leads to determine the temporary lead tip position of the adjacent leads, It is characterized in that the above-described scanning is repeated to sequentially determine the provisional lead tip positions of adjacent leads.

According to a third aspect of the present invention, in the image processing apparatus, the difference between the number of temporary lead tip positions obtained by the lead tip temporary position measuring means and a predetermined number of leads of the semiconductor device is fixed. The above case is characterized in that an erroneous recognition detecting means for recognizing that the detected temporary lead tip position is incorrect is provided.

According to a fourth aspect of the present invention, the image processing apparatus further scans in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of a part recognized as an object is less than a predetermined value. If it is determined that the read is not a read, and if it is determined that the read is not a read, the scan is performed in the horizontal scanning direction without calculating the provisional position of the leading end of the lead.

According to a fifth aspect of the present invention, a temporary lead tip position is measured for each lead of the semiconductor device based on the binarized image information, and the temporary lead tip position and a grayscale image of a surrounding area are measured. The true lead tip position is calculated for each lead, and the center of gravity and the inclination of the semiconductor device are calculated based on the true lead tip position calculated by the true lead tip position calculating means.

According to a sixth aspect of the present invention, in the image processing method, the semiconductor device is further mounted in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and a vertical scanning direction perpendicular to the horizontal scanning direction is provided from the temporary lead tip position. Then, by moving a predetermined number of pixels in the direction of the root of the lead and scanning in the horizontal scanning direction to detect the adjacent lead, and then moving in the vertical scanning direction and the leading end direction of the lead, the adjacent lead is moved. By calculating the temporary lead tip position and repeating the above scanning,
It is characterized in that the temporary lead tip positions of adjacent leads are sequentially obtained.

In the image processing method according to the present invention, the difference between the number of temporary lead tip positions obtained by the lead tip temporary position measuring means and a predetermined number of leads of the semiconductor device may be a fixed value. In the above case, the detected temporary lead tip position is recognized as being incorrect.

In the image processing method according to the present invention, scanning is performed in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of a portion recognized as an object is equal to or smaller than a predetermined value, the reading is performed. If it is determined that the lead is not a lead, and if it is determined that the lead is not a lead, the lead tip temporary position is not calculated,
Further, scanning is performed in the horizontal scanning direction.

[0018]

In the image processing apparatus according to the first aspect, the lead tip temporary position measuring means measures the temporary lead tip position for each lead of the semiconductor device based on the binarized image information. Further, the true lead tip position calculating means calculates the true lead tip position for each lead from the grayscale image of the temporary lead tip position and its surrounding area.

Therefore, from the provisional lead tip position measured based on the binarized image information, a true lead tip position corrected in consideration of the gray value can be obtained.

Further, the center of gravity calculating means calculates the center of gravity and the inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. Therefore, even when the shape of the semiconductor device is close to a square, the position and orientation of the semiconductor device can be measured with high accuracy.

In the image processing apparatus according to the second aspect, the semiconductor device is further mounted such that the arrangement direction of the leads substantially coincides with the horizontal scanning direction, and the semiconductor device is positioned perpendicular to the horizontal scanning direction from the temporary lead tip position. In the vertical scanning direction and in the root direction of the lead, move by a predetermined pixel, detect an adjacent lead by scanning in the horizontal scanning direction, and then move in the vertical scanning direction and in the tip direction of the lead, The temporary lead tip positions of the adjacent leads are obtained, and the above-described scanning is repeated, thereby sequentially obtaining the temporary lead tip positions of the adjacent leads. As described above, after temporarily moving from the provisional lead tip position to the root direction of the lead, the adjacent lead is detected, and by moving from the root direction of the adjacent lead to the tip direction, the adjacent lead can be more quickly and reliably connected. The temporary lead tip position can be obtained.

In the image processing apparatus according to the third aspect, the erroneous recognition detecting means includes the number of provisional lead tip positions obtained by the lead tip provisional position measuring means, and a predetermined number of leads of the semiconductor device. If the difference is equal to or greater than a certain value, it is recognized that the detected temporary lead tip position is incorrect. Therefore, it is possible to prevent erroneous detection of dust or the like having substantially the same size as the lead.

In the image processing apparatus according to the fourth aspect, scanning is performed in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of a portion recognized as an object is less than a predetermined value, Judge that it is not a lead. Therefore, noise smaller than the lead is not erroneously detected. When it is determined that the read is not a lead, the scan in the horizontal scanning direction is performed without calculating the temporary position of the lead tip. Therefore, it is possible to prevent the lead tip temporary position from being erroneously calculated.

In the image processing method according to the fifth aspect, a provisional lead tip position is measured for each lead of the semiconductor device based on the binarized image information, and a gradation image of the provisional lead tip position and a surrounding area is measured. , The true lead tip position is calculated for each lead.

Therefore, from the provisional lead tip position measured based on the binarized image information, a true lead tip position corrected in consideration of the gray value can be obtained.

The center of gravity calculating means calculates the center of gravity and the inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. Therefore, even when the shape of the semiconductor device is close to a square, the position and orientation of the semiconductor device can be measured with high accuracy.

[0027] In the image processing method according to the sixth aspect, the semiconductor device is further mounted in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and a vertical scanning perpendicular to the horizontal scanning direction is performed from the leading end position of the temporary lead. In the direction and the root direction of the lead, the adjacent lead is detected by moving by a predetermined pixel, scanning in the horizontal scanning direction, and then moving in the vertical scanning direction and the tip direction of the lead, thereby detecting the adjacent lead. It is characterized in that a temporary lead tip position of a lead is obtained, and the above-described scanning is repeated to sequentially obtain a temporary lead tip position of an adjacent lead. As described above, after temporarily moving from the provisional lead tip position to the root direction of the lead, the adjacent lead is detected, and by moving from the root direction of the adjacent lead to the tip direction, the adjacent lead can be more quickly and reliably connected. The temporary lead tip position can be obtained.

According to the image processing method of the present invention, the difference between the number of temporary lead tip positions obtained by the temporary lead tip position measuring means and the predetermined number of leads of the semiconductor device is constant. When the value is equal to or larger than the value, the detected temporary lead tip position is recognized as being erroneous. Therefore, it is possible to prevent erroneous detection of dust or the like having substantially the same size as the lead.

In the image processing method according to the eighth aspect, scanning is performed in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of a part recognized as an object is smaller than a predetermined value, If it is determined that the read is not a read, and if it is determined that the read is not a read, the scan is performed in the horizontal scanning direction without calculating the provisional position of the leading end of the lead.

Therefore, there is no possibility of erroneously detecting noise or the like smaller than the lead. When it is determined that the read is not a lead, the scan in the horizontal scanning direction is performed without calculating the temporary position of the lead tip. Therefore, it is possible to prevent the lead tip temporary position from being erroneously calculated.

[0031]

An embodiment of the present invention will be described with reference to the drawings. First, FIG. 1 shows an overall configuration diagram of an image processing apparatus according to an embodiment of the present invention. This apparatus includes an imaging unit 21, a lead tip temporary position measuring unit 23, a true lead tip position calculating unit 25, an erroneous recognition detecting unit 29, and a center-of-gravity calculating unit 27. An outline of the relationship between the means will be described.

The image information obtained by the imaging means 21 is 2
It is converted to the digitized image information, and the lead tip temporary position measuring means 23
Given to. The lead tip temporary position measuring means 23 measures the temporary lead tip position for each lead of the semiconductor device while removing noise. When measuring the provisional lead tip position, the erroneous recognition detection means 29 determines that the difference between the number of provisional lead tip positions obtained by the lead tip provisional position measurement means and the predetermined number of semiconductor device leads is constant. If the value is equal to or larger than the value, it is recognized that the detected temporary lead tip position is incorrect.

If it is recognized that the detected temporary lead tip position is not erroneous, the true lead tip position calculating means 25
Calculates the true lead tip position for each lead from the temporary lead tip position and the grayscale image of the surrounding area. The center of gravity calculating means 27 calculates the center of gravity and the inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means.

FIG. 2 shows an image processing apparatus 13 in which each function of FIG. 1 is realized using a CPU. The image processing device 13 includes a camera 1 serving as an imaging unit, an image input unit 2, an image memory 3, an image output unit 4, a video monitor 5, a timing control unit 6, a character memory 7, an address / data bus 8, a CPU 9, a RO
M10, RAM 11 and I / O 12 are provided.

The camera 1 sends the captured image to the image input unit 2 as an analog video signal. The image input unit 2
A / D conversion of the input video signal sends the grayscale image data and the binary image data to the image memory 3.

The image memory 3 stores the input image data in units of one pixel. The character memory 7 stores font data for characters to be displayed on the video monitor 5. The image output unit 4 performs D / A conversion on image data composed of digital signals output from the image memory 3 and the character memory 7 to convert them into video signals for one screen.
Send to video monitor 5. The video monitor 5 displays an input image, a calculation result, and the like based on the video signal output from the image output unit 4.

On the other hand, the image memory 3, the character memory 7, the CPU 9, the ROM 10, the RAM
A microcomputer is configured by connecting the I / O 12. In this microcomputer, the CPU 9 executes a program stored in the ROM 10. As a result, various image processing such as measurement of the position and orientation of the IC is executed.

The timing control unit 6 outputs a timing signal for controlling the input and output of data in the image input unit 2, the image memory 3, the image output unit 4, and the character memory 7 in cooperation with the CPU 9.

The operation of this device will be described with reference to the flowcharts of FIGS. 3, 4, 5 and 6. First, Q
About the lead at the top of FP (quad flat package) 51,
The case of performing the processing will be described. The CPU 9 converts the image data stored in the image memory 3 into an arrow SC in units of one pixel.
Scanning is sequentially performed in the direction 1 (horizontal scanning direction, see FIG. 3A) (step S1 in FIG. 4). Point P where the binary data becomes "1"
Is detected (see FIG. 3A), a predetermined area including the point P (here, 3 * 2) is set, and within this area,
It is determined how many pixels have binary data “1”. If the number of pixels whose binary data is “1” is equal to or less than a predetermined number (here, three pixels), scanning is continued assuming that noise is present (step S2 in FIG. 4). Note that the size of the predetermined area and the predetermined value may be determined based on the size of one pixel, the size of noise, and the like.

If it is determined in step S2 that the area is not noise, the point is set as the provisional lead tip position (point P1 (X1, Y1) in FIG. 3B) in the lead (step S3 in FIG. 4). 3B and 3C show a method of measuring a temporary lead tip position of a lead adjacent to the lead.
This will be described with reference to FIG.

First, the binary data of the lower three pixels of the point P1 is checked. In this case, the cases of FIGS. 3C to 3H are considered depending on the quantization error of the binarization and the scanning direction. Here, since it corresponds to FIG. 3F, it moves in the lower right direction (inward of the lead). E and F in FIG. 8 show the case where scanning is performed in the direction of arrow SC1, and G and H in FIG.
This is a case where scanning is performed in a direction opposite to the direction of 1. In the cases of FIGS. C and D, there are both cases. By performing such correction movement, the QFP 51 can be reliably moved to the inside of the lead even if the position and attitude of the QFP 51 are shifted.

The number of pixels to be corrected and moved is one pixel in this embodiment, but may be determined based on the width of the lead and the size of one pixel.

Next, returning to FIG. B, the image is moved by n pixels in the direction of arrow SC2. This point is designated as P2. The value of the n pixels to be moved here is the QFP
Based on the 51 data (the maximum allowable inter-lead pitch and the maximum inclination of the posture when conveyed), it may be determined that the point P2 moved by n pixels is present inside the lead to be measured. . At the point during tracking, the Y coordinate is y = (y1
The position where (+ y2) / 2 is set as the point P3, and the point P2 moves to the point P3. As described above, after the predetermined number of pixels have been moved, adjacent leads are detected.

The detection of adjacent leads is performed as follows. Scan in the direction opposite to arrow SC1 to
A point at which the value data once becomes 0 and then becomes 1 again is defined as a point P4. Here, the correction movement (corresponding to FIG. 3C in this case) is performed again, and then scanning is performed in the direction (vertical scanning direction) opposite to the arrow SC2 which is the leading end direction of the lead, and the temporary lead leading end position of the lead is read. P
Find 5 In this manner, the temporary lead tip position P5 of the adjacent lead is obtained.

After that, it moves by n pixels in the direction of arrow SC2. This point is designated as P6. At the point during tracking, the Y coordinate is
A position where y = (y5 + y6) / 2 is set as a point P7, and a point P6
From the point P7. By repeating such an operation, the position of the provisional lead tip is sequentially detected. In addition, even if the binary data is scanned once in the direction opposite to the direction of the arrow SC1 and the pixel data is moved twice as much as the pitch between the leads of the QFP 51, the point which becomes 1 again is not found. It is determined that there is no lead to be read, and it is only necessary to move to the point P3 and scan from that position in the direction of the arrow SC1. In addition,
In the present embodiment, the case where the first temporary lead tip position does not start from the lead located at the end has been described, but if the first temporary lead tip position starts from the lead located at the end, the arrow SC1 Scanning may be performed in either the direction or the opposite direction. Next, the CPU 9
Compares the number of the provisional lead tip positions obtained for all the leads on the upper side with the number of leads on the upper surface of the QFP 51 stored in the ROM 10 or the RAM 11 (step S4 in FIG. 4). If the difference between the two is equal to or greater than a certain value, it is recognized that the detected temporary lead tip position is not a lead. Thus, it is possible to prevent dust or the like having substantially the same size as the lead from being erroneously detected as the lead.

The above operation is repeated for the other three surfaces, and the leading end positions of the temporary leads are obtained for all the upper leads.

Thereafter, the true lead tip position is calculated based on the provisional lead tip position by using the grayscale image clipping (step S5 in FIG. 4). Hereinafter, a method of extracting a grayscale image will be described.

First, as shown in FIG. 5, a pixel 16 corresponding to the position of the leading end of the tentative lead is set as a pixel of interest.
, A rectangular mask 17 including eight neighboring pixels
Is set, and it is determined whether or not each of the eight neighboring pixels satisfies a predetermined condition (hereinafter, referred to as a “first condition”) using the binarization threshold value TH. It is determined whether or not the pixel constitutes the interior, and if the first condition is satisfied, the pixel is sequentially extracted as a set element. In the same figure, each cell indicates a pixel.

FIG. 6 shows the position of each pixel in the mask 17 by XY coordinate values.
_i , Y _i ), the XY coordinate values of each pixel in the vicinity of 8 are represented as shown in FIG.

Assuming that the gray level of the neighboring pixel 18 at the pixel position (X _i -1, Y _i -1) is F (X _i -1, Y _i -1), the above first condition is given by the following equation: Given by

F (X _i −1, Y _i −1) * δ (X _i −1, Y _i −1)> TH In the above formula, δ (X _i −1, Y _i −1) is This is a function indicating whether or not the neighboring pixel 18 is already a set element. If the function is not already a set element, the value of this function is “0”.
The value of this function is “1”.

As is apparent from this equation, when the image portion of the object is brighter than the background image portion, the gray level of the neighboring pixel 18 at the pixel position (X _i -1, Y _i -1) is 2
If it is larger than the threshold value TH and the neighboring pixels 18 are not yet a set element, the first condition is satisfied. If the gray level is equal to or less than the binarization threshold value TH, or if the neighboring pixels 18 are already an aggregate element, the first condition is not satisfied.

On the other hand, when the image portion of the object is darker than the background image portion, the gray level of the neighboring pixel 18 at the pixel position (X _i -1, Y _i -1) is smaller than the binarization threshold TH and is smaller than the threshold TH. If the neighboring pixel 18 is not yet a set element, the first
The condition will be satisfied. If the gray level is equal to or greater than the binarization threshold TH, or if the neighboring pixels 18 are already an aggregate element, the first condition is not satisfied.

The same applies to the other eight neighboring pixels, and the description is omitted here.

In this way, it is sequentially determined whether each of the eight neighboring pixels satisfies the first condition or not. Pixels satisfying the first condition are registered in the RAM 11 in addition to the set elements. In FIG. 5, each pixel marked with a cross is a pixel registered as an aggregate element.

When the condition determination of each of the eight neighboring pixels is completed, each pixel newly added to the set element (in this example, 8
The neighboring pixels) are sequentially designated as the pixel of interest, and similar masks 17 are sequentially set around the pixel of interest as shown by the broken line in FIG. Will be done. This procedure is repeated until the newly added set element is exhausted.

After the pixels constituting the inside of the image are extracted from the image portion 15a of the object as described above, the pixels constituting the boundary of the image are extracted.

FIG. 7 shows the extraction method. In FIG. 7, each pixel marked with x is a pixel registered as a pixel (collective element) constituting the inside of the image.

Here, each set element is sequentially set as a pixel of interest,
After setting a rectangular mask 17 similar to the above including the eight neighboring pixels as shown in FIG. 4 around each pixel of interest, the background gray level LV is set for each of the eight neighboring pixels.
By determining whether or not a predetermined condition (hereinafter, referred to as a “second condition”) using the gray level LV2 of the pixel of interest 1 and the pixel of interest is satisfied, whether or not each pixel is a pixel constituting a boundary of an image is determined. Then, it is determined whether or not the pixel is a pixel constituting an image portion of an adjacent object, and if the second condition is satisfied, the pixel is sequentially extracted as a set element.

The pixel position (X _i −
The gray level of the pixel of (1, Y _i −1) is represented by F (X _i −1, Y _i).
If -1), the above-mentioned second condition is given by the following equation.

F (X _i −1, Y _i −1) * δ (X _i −1, Y _i −1)> LV1... F (X _i −1, Y _i −1) * δ (X _i −1, Y _i −1) <LV2 In the above equation, δ (X _i −1, Y _i −1) is a function indicating whether or not the pixel is already a set element. If it is a set element, the value of this function is "0". If it is not a set element, the value of this function is "1". As is apparent from this equation, when the image portion of the object is brighter than the image portion of the background, the gray level of the pixel at the pixel position (X _i -1, Y _i -1) is larger than the gray level LV1 of the background. If the pixel is not yet a set element and is smaller than the gray level LV2 of the target pixel, the second condition is satisfied. If the gray level is equal to or lower than the background gray level LV1, or if the pixel is already an aggregate element, or if the gray level is equal to or higher than the gray level LV2 of the target pixel, the second condition is not satisfied. .

On the other hand, if the image portion of the object is darker than the background image portion, the gray level of the pixel at the pixel position (X _i -1, Y _i -1) is smaller than the gray level LV1 of the background and the pixel is If it is not a set element yet and is larger than the gray level LV2 of the pixel of interest, the second condition is satisfied. If the gray level is equal to or higher than the background gray level LV1, or if the pixel is already an aggregate element, or if the gray level is lower than the gray level LV of the pixel of interest, the second condition is not satisfied. Become.

The same applies to the other eight neighboring pixels, and the description is omitted here.

In this way, it is sequentially determined whether each of the eight neighboring pixels satisfies the second condition, and the pixels satisfying the second condition are registered in the RAM 11 in addition to the set elements. In the figure, each pixel marked with a broken line x is a pixel newly registered as an aggregate element.

When the condition determination of each of the eight neighboring pixels is completed, each pixel which is the next set element and each pixel which is a newly added set element are sequentially set as a pixel of interest. Are sequentially set, and the condition determination is similarly performed for each of the eight neighboring pixels. This procedure is repeated until the newly added set element is exhausted.

FIGS. 8 and 9 show a measurement procedure by the CPU 9 based on the above principle.

In order to simplify the description, FIGS.
The measurement procedure will be specifically described for the grayscale image 21 shown in FIG. In the figure, each cell is a pixel, and in the image portion 20 of the object, each pixel constituting the inside of the A to E images is
In addition, a to 1 indicate each pixel constituting the boundary.

Assuming that the first pixel of interest is A, in step 1 (shown as “ST1” in the figure) of FIG.
U9 registers the coordinates (X ₀ , Y ₀ ) of the target pixel A in the storage areas X (0), Y (0) of the RAM 11 as the 0th set element, and initializes the counters n, i of the CPU 9 to zero. Set.

After setting the count value of the counter n to another counter en of the CPU 9 in the next step 2, the CPU 9 sets the target pixel A in the next step 3.
, The coordinates (X _i −
It is determined whether the first condition is satisfied for the pixel of (1, Y _i -1).

The counters n and en are for counting the number of set elements, and the counter i is for counting the order of the pixel of interest.

If the determination in step 3 is "YES", the process proceeds to step 4, where the counter n is incremented by 1 and the coordinates (X _i -1, Y _i -1) of the pixel are set to the first. Storage area X (1), Y of RAM 11 as an aggregate element
In the specific example shown in FIG. 10, since the first neighboring pixel 22 does not satisfy the first condition, the step 4 is skipped, the process proceeds to the step 5, and the CPU 9 Determines whether the second neighboring pixel 23 with respect to the pixel of interest A, that is, the pixel at the coordinates (X _i , Y _i -1) satisfies the first condition.

If the determination in step 5 is "YES", the process proceeds to step 6, where the counter n is incremented by 1, and the coordinates (X _i , Y _i -1) of the pixel are stored in the RAM 11 as a set element. However, in the specific example shown in the drawing, since the second neighboring pixel 23 does not satisfy the first condition, this step 4 is also skipped.

Similarly, in the procedure up to steps 7 and 8, it is determined whether or not the third to eighth neighboring pixels satisfy the first condition. If the first condition is satisfied, the counter n is added. Registration of a set element is performed.

In the specific example shown in the figure, the neighboring pixels B and C satisfy the first condition, and the coordinates of the pixel B are stored in the storage areas X (1) and Y (1) of the RAM 11 as the first set element.
And the coordinates of the pixel C are registered in the storage areas X (2) and Y (2) as the second set element. The counter n is incremented each time it is registered. As a result, the value of the counter n is "2".

Next, at step 9, the counter i is incremented by 1, and at the next step 10, the value of the counter i (in this case, "1") is compared with the value of the counter en (in this case, "0"). In this case, the value of the counter i is
n, the determination in step 10 is “YE
S ", the value of the counter i is set to a value obtained by adding 1 to the value of the counter en (in this case," 1 ") in the next step 11, and the pixel B is designated as the target pixel. The coordinates of the pixel B are given by (X _i , Y _i ).

In the next step 12, it is determined whether or not the value of the counter n (in this case, "2") matches the value of the counter en (in this case, "0"). Is "NO", the process returns to step 2 and rewrites the value of the counter en to the value of the counter n.
In this case, since the value of the counter n is “2”, the value of the counter en is rewritten to “2”.

Hereinafter, in steps 3 to 8,
The CPU 9 determines whether or not the first to eighth neighboring pixels with respect to the pixel of interest B satisfy the first condition. In the case of the specific example in the drawing, the neighboring pixels D, E, F, and G are the first. The condition is satisfied, and the coordinates of the pixel D are stored in the storage areas X (3) and Y (3) of the RAM 11 as the third set element, and the coordinates of the pixel E are stored in the storage area X as the fourth set element.
(4), Y (4), the storage area X (5), Y (5) as the fifth set element of the coordinates of the pixel F, and the storage area X (6) as the sixth set element of the coordinates of the pixel G. , Y (6)
Are registered respectively. The counter n is incremented each time it is registered. As a result, the value of the counter n is "6".

Next, at step 9, the counter i is incremented by one, and at the next step 10, the value of the counter i (in this case "2") is compared with the value of the counter en (in this case "2"). In this case, the value of the counter i and the counter e
Since the value of n is equal, the determination in step 10 is “NO”
Then, the process returns to step 3, and the CPU 9 determines in the following steps 3 to 8 whether or not the first to eighth neighboring pixels for the target pixel C satisfy the first condition.

In the case of the specific example shown in the figure, since there is no longer any pixel that satisfies the first condition, there is no newly registered set element, and the value of the counter n remains "6".

Next, at step 9, the counter i is incremented by one, and at the next step 10, the value of the counter i (in this case "3") is compared with the value of the counter en (in this case "2"). In this case, the value of the counter i is
n, the determination in step 10 is “YE
S ", the value of the counter i is set to a value obtained by adding 1 to the value of the counter en (in this case," 3 ") in the next step 11, and the pixel D is designated as the target pixel. The coordinates of the pixel D are given by (X ₃ , Y ₃ ).

The next step 12 is to determine whether or not the value of the counter n (in this case, "6") matches the value of the counter en (in this case, "2"). Is "NO", the process returns to step 2 and rewrites the value of the counter en to the value of the counter n.
In this case, since the value of the counter n is “6”, the value of the counter en is rewritten to “6”.

Hereinafter, the CPU 9 performs the first
It is determined whether the first condition is satisfied for the eighth to eighth neighboring pixels.
It is determined whether or not the first pixel satisfies the first condition. In the case of the specific example shown in FIG. 8, there is no pixel that satisfies the first condition. , Step 12 becomes “YES”, and the flow proceeds to the second condition determination procedure shown in FIG.

First, at step 13, the CPU 9 sets the counter i
Is set to "0", and in the following step 14, the counter en
Is set to the value of the counter n (in this case, “6”), and in the next step 15, the first neighboring pixel 22 with respect to the target pixel A at the position of the coordinates (X ₀ , Y ₀ ), that is, the coordinates (X _i -1, to determine whether the second condition is satisfied for each pixel of the Y _i -1).

If the determination in step 15 is "YES", the process proceeds to step 16, where the counter n is incremented by 1, and the coordinates (X _i -1, Y _i -1) of the pixel are set to the seventh position. Storage area X (7) of RAM 11 as an aggregate element,
In the specific example shown in FIG. 11, since the first neighboring pixel 22 does not satisfy the second condition, step 16 is skipped and the process proceeds to step 17, and The CPU 9 determines whether or not the second neighboring pixel with respect to the pixel of interest A, that is, the pixel a at the coordinates (X _i , Y _i -1) satisfies the second condition.

In the case of the specific example shown in the figure, since the determination in step 17 is "YES", the flow proceeds to step 18, and
The counter n is incremented by 1 (n = 7), and the coordinates (X _i , Y _i -1) of the pixel a are set as a set element in the RAM
In the storage areas X (7) and Y (7).

Similarly, in the procedure up to steps 19 and 20, it is determined whether or not the third to eighth neighboring pixels satisfy the first condition. If the first condition is satisfied, the counter n is added. Registration of a set element is performed.

In the specific example shown in the figure, neighboring pixels b, c,
d satisfies the second condition, and the coordinates of the pixel b are stored in the storage areas X (8), Y
(8), the coordinates of the pixel c are stored in the storage areas X (9) and Y (9) as the ninth set element, and the coordinates of the pixel d are set to 10
As storage elements X (10) and Y (1
0) are registered. The counter n is incremented each time it is registered. As a result, the value of the counter n is "10".

Next, at step 21, the counter i is incremented by one, and at the next step 22, the value of the counter i (in this case "1") is compared with the value of the counter en (in this case "6"). In this case, since the value of the counter i is smaller than the value of the counter en, the determination in step 22 is “N”.
O "and returns to step 15, and then the CPU 9
In the following steps 15 to 20, it is determined whether or not the first to eighth neighboring pixels for the target pixel B satisfy the second condition. In the case of the specific example shown in the figure, the neighboring pixel e satisfies the second condition, and the coordinates of the pixel e are stored in the storage area X (1
1) and Y (11) are registered. At this stage, the value of the counter n is "11".

Next, at step 21, the counter i is incremented by 1. At step 10, the value of the counter i (in this case, "2") is compared with the value of the counter en (in this case, "6"). In this case, the value of the counter i is smaller than the value of the counter en.
O "and returns to step 3. Next, in step 3 to step 8, the CPU 9 determines whether or not the first to eighth neighboring pixels for the target pixel C satisfy the second condition. Subsequently, it is similarly determined whether or not the first to eighth neighboring pixels for the target pixels D to G satisfy the second condition.

When the condition determination for each of the target pixels C to G is completed in this way and the content of the counter i is incremented by 1 in step 21, i = 7, and the determination in step 22 is "YES". Becomes In the next step 23, the counter i is set to a value obtained by adding 1 to the value of the counter en (in this case, "7"), and the pixel a is designated as the target pixel. Note that the coordinates of the pixel a are given by (X ₇ , Y ₇ ).

In the next step 24, it is determined whether or not the value of the counter n (in this case, "17") matches the value of the counter en (in this case, "6"). In this case, the determination is made. Is "NO", the process returns to the step 14, and the value of the counter en is rewritten to the value of the counter n.
In this case, since the value of the counter n is “17”, the value of the counter en is rewritten to “17”.

Hereinafter, the CPU 9 performs the first
It is determined whether or not the second condition is satisfied for the eighth to eighth neighboring pixels. Subsequently, the pixels b to l are sequentially focused to determine whether the second condition is similarly satisfied for the eight neighboring pixels. Eventually, in the specific example of the figure, since there is no pixel that satisfies the second condition, when the condition determination for the pixel 1 is completed, the determination in step 24 becomes “YES”, and the determination of the second condition is performed. When the procedure is completed, the process proceeds to step S25.

Using the set of pixels extracted in this manner as a measurement area, in step 25, the CPU 9 first calculates the area S of the measurement area by performing the operation of the following equation. Is calculated (X _G , Y _G ).

[0094]

(Equation 1)

[0095]

(Equation 2)

[0096]

(Equation 3)

By utilizing such a technique, 2
An accurate lead shape can be obtained from the schematic shape of the lead obtained from the value image. Further, by setting a set of pixels as a measurement area to a predetermined number of pixels from the lead end pixels constituting each lead, a true lead end position which is the center of gravity of the lead end can be obtained.

The relationship between the provisional lead tip position and the true lead tip position is as follows. In contrast, when a grayscale image as shown in FIG. 13A is obtained, it is binarized by setting the grayscale level 9 as a threshold value (the grayscale level 9 or more is set to “1”), and the binary image shown in FIG. Value image can be obtained. When this binary image is scanned in the horizontal scanning direction (the direction of the arrow SC1), the point P1 where the binary data becomes "1" first becomes the provisional lead tip position in the lead. With the point P1 as the temporary lead tip position, a gray image is cut out in the range of the window. In this case, the pixels 31a1, 31b1, 31 as shown in FIG.
c1,31d1,31e1,31a2,31b2,31c2,31d2,31e2,31a3,31b3,31
The average position of each of the loads c3, 31d3, and 31e3 is obtained by the above equation, and the obtained point is the true lead end position.

In this embodiment, the predetermined number of pixels from the lead end pixel is set to three, but may be larger than that.

With the above processing, the true lead tip position is calculated based on the temporary lead tip position obtained from the binary image. From the calculated true lead tip position, the position and orientation of the QFP 51 are obtained as follows.

When obtaining the position of the QFP 51, the average of all true lead tip positions may be obtained. In this case, since the true lead tip position obtained is higher in accuracy than the binary image data, the position of the QFP 51 can be obtained accurately.

When obtaining the attitude of the QFP 51, the average point of the true lead tip position is determined for each of the upper, lower, left and right leads, and a straight line connecting the opposing points is determined. Then, the inclination of the straight line may be used as the attitude of the QFP 51. In particular,
As shown in FIG. 7C, the average of the X coordinate of the true lead tip position of the upper lead is defined as X1, and similarly, the average of the Y coordinate Y
Find 1 Similarly, average X2, Y2 of the lower lead,
Average left lead X3, Y3, average right lead X4, Y
Ask for 4. Then, the points (X1, Y1) and (X2, Y
The straight line connecting 2) may be obtained, and the inclination of the straight line may be used as the attitude of the QFP 51. Note that point (X3, Y3) and point (X4,
The same applies to the case where a straight line connecting Y4) is obtained.

If all leads could not be detected due to missing leads or bent pins, etc.
As shown in A, from the linear equation obtained by the least square method,
By calculating the points P100 (X100, Y100) to P103 (X103, Y103) as follows, the position and orientation of the QFP 51 can be obtained.

Position X = (X100 + X101 + X102 + X103) / 4 Position Y = (Y100 + Y101 + Y102 + Y103) / 4 Attitude θ = (θ100 + θ101 + θ102 + θ103) / 4 where θ100, θ101, θ102 and θ103 are as follows: Is given by

Θ100 = TAN- ¹ ｛(Y100-Y101) / (X100-X101)｝ θ101 = TAN- ¹ ｛(Y101-Y102) / (X101-X102)｝ θ102 = TAN- ¹ ｛(Y102-Y103) / (X102-X103)｝ θ103 = TAN ^-1 {(Y103-Y100) / (X103-X100)} As shown in FIG. 12B, the light reflected by irradiating the illumination light from the light source to the QFP 51 is converted into a binary value. Figure C
A binary image as shown in FIG. In such a case,
An arrow SC2 (FIG. 3)
When the pixel is moved by n pixels in the direction (see B), the point P2 is defined once after it once goes out of the lead. Therefore,
As described above, when the light beam once goes out of the lead before moving by n pixels, the point that once goes out of the lead may be set to P2. In this case as well, the position of the point P3 may be the average of each of the points P2 and P1, as in the method described above.

In this embodiment, Q is used as the IC.
Although the description has been made using the FP, other forms of IC, for example, 2
It may be used when measuring an IC having a lead in a direction.

[0107]

In the image processing apparatus according to the first aspect, the lead tip temporary position measuring means measures the temporary lead tip position for each lead of the semiconductor device based on the binarized image information. In addition, the true lead tip position calculating means includes:
The true lead tip position is calculated for each lead from the temporary lead tip position and the grayscale image of the surrounding area.

Therefore, from the provisional lead tip position measured based on the binarized image information, a true lead tip position corrected in consideration of the gray value can be obtained.

Further, the center of gravity calculating means calculates the center of gravity and the inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. Therefore, it is possible to provide an image processing apparatus capable of measuring the position and orientation of the semiconductor device with high accuracy even when the shape of the semiconductor device is close to a square.

In the image processing apparatus according to the second aspect, the semiconductor device is further mounted such that the arrangement direction of the leads substantially coincides with the horizontal scanning direction, and is perpendicular to the horizontal scanning direction from the temporary lead tip position. In the vertical scanning direction and in the root direction of the lead, move by a predetermined pixel, detect the adjacent lead by scanning in the horizontal scanning direction, and then move in the vertical scanning direction and in the tip direction of the lead, The temporary lead tip positions of the adjacent leads are obtained, and the above-described scanning is repeated, thereby sequentially obtaining the temporary lead tip positions of the adjacent leads. As described above, after temporarily moving from the provisional lead tip position to the root direction of the lead, the adjacent lead is detected, and by moving from the root direction of the adjacent lead to the tip direction, the adjacent lead can be more quickly and reliably connected. The temporary lead tip position can be obtained.

In the image processing apparatus according to the third aspect, the erroneous recognition detecting means includes the number of provisional lead tip positions obtained by the lead tip provisional position measuring means, and the number of leads of the semiconductor device given in advance. If the difference is equal to or greater than a certain value, it is recognized that the detected temporary lead tip position is incorrect. Therefore, it is possible to prevent erroneous detection of dust or the like having substantially the same size as the lead. Accordingly, it is possible to provide an image processing apparatus capable of measuring the position and orientation of the semiconductor device with high accuracy even when dust is present around the semiconductor.

In the image processing apparatus according to the fourth aspect, scanning is performed in the horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of the part recognized as an object is less than a predetermined value, Judge that it is not a lead. Therefore, noise smaller than the lead is not erroneously detected. When it is determined that the read is not a lead, the scan in the horizontal scanning direction is performed without calculating the temporary position of the lead tip. Therefore, it is possible to prevent the erroneous calculation of the temporary position of the leading end of the lead, and it is possible to shorten the detection time of the temporary position of the leading end as a whole.

Thus, it is possible to provide an image processing apparatus capable of measuring the position and orientation of the semiconductor device with high accuracy even when image noise is present.

In the image processing method according to the fifth aspect, the provisional lead tip position is measured for each lead of the semiconductor device on the basis of the binarized image information, and the gradation image of the provisional lead tip position and the surrounding area is measured. , The true lead tip position is calculated for each lead.

Therefore, from the provisional lead tip position measured based on the binarized image information, a true lead tip position corrected in consideration of the gray value can be obtained.

Further, the center of gravity calculating means calculates the center of gravity and the inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. Therefore, even when the shape of the semiconductor device is close to a square, the position and orientation of the semiconductor device can be measured with high accuracy.

[0117] In the image processing method according to the sixth aspect, the semiconductor device is further mounted in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and the vertical scanning perpendicular to the horizontal scanning direction is performed from the temporary lead tip position. In the direction and the root direction of the lead, the adjacent lead is detected by moving by a predetermined pixel, scanning in the horizontal scanning direction, and then moving in the vertical scanning direction and the tip direction of the lead, thereby detecting the adjacent lead. It is characterized in that a temporary lead tip position of a lead is obtained, and the above-described scanning is repeated to sequentially obtain a temporary lead tip position of an adjacent lead. As described above, after temporarily moving from the provisional lead tip position to the root direction of the lead, the adjacent lead is detected, and by moving from the root direction of the adjacent lead to the tip direction, the adjacent lead can be more quickly and reliably connected. The temporary lead tip position can be obtained.

According to the image processing method of the present invention, the difference between the number of temporary lead tip positions obtained by the temporary lead tip position measuring means and the predetermined number of leads of the semiconductor device is constant. When the value is equal to or larger than the value, the detected temporary lead tip position is recognized as being erroneous. Therefore, it is possible to prevent erroneous detection of dust or the like having substantially the same size as the lead.

Thus, it is possible to provide an image processing apparatus capable of measuring the position and orientation of a semiconductor device with high accuracy even when dust is present around the semiconductor.

In the image processing method according to the eighth aspect, scanning is performed in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and when the number of pixels of a part recognized as an object is less than a predetermined value, If it is determined that the read is not a read, and if it is determined that the read is not a read, the calculation of the provisional position of the leading end of the lead is not performed, and scanning is performed in the horizontal scanning direction.

Therefore, a noise smaller than the lead is not erroneously detected. When it is determined that the read is not a lead, the scan in the horizontal scanning direction is performed without calculating the temporary position of the lead tip. Therefore, it is possible to prevent the erroneous calculation of the temporary position of the leading end of the lead, and it is possible to shorten the detection time of the temporary position of the leading end as a whole. Accordingly, it is possible to provide an image processing apparatus capable of measuring the position and orientation of the semiconductor device with high accuracy even when image noise is present.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image processing apparatus according to an embodiment of the present invention.

FIG. 2 is a diagram illustrating an example of a hardware configuration in which each function of the image processing apparatus is implemented using a CPU.

FIG. 3 is a diagram illustrating an operation of obtaining a temporary lead tip position.

FIG. 4 is a flowchart showing the entire operation of the CPU 9;

FIG. 5 is a diagram illustrating a method of calculating a true lead tip position.

FIG. 6 is an explanatory diagram showing each pixel position in a mask.

FIG. 7 is a principle explanatory diagram showing a method of extracting a set of pixels of an object.

FIG. 8 is a flowchart for calculating a true lead tip position by a CPU 9;

FIG. 9 is a flowchart for calculating a true lead tip position by a CPU 9;

FIG. 10 is a principle explanatory diagram showing a specific example of a grayscale image and a method of extracting a set of pixels.

FIG. 11 is a principle explanatory diagram showing a specific example of a grayscale image and a method of extracting a set of pixels.

FIG. 12 is a diagram showing a method for obtaining a straight line equation by the least square method.

FIG. 13 is a diagram for explaining a true lead tip position and a provisional lead tip position, and an explanatory diagram for obtaining a linear equation from the true lead tip position.

[Explanation of symbols]

 21 ・ ・ ・ Imaging means 23 ・ ・ ・ Temporary lead tip position measuring means 25 ・ ・ ・ True lead tip position calculating means 27 ・ ・ ・ Center of gravity calculating means 29 ・ ・ ・ Error recognition detecting means

Continued on the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G01B 11/00-11/30 102 G01N 21/84-21/958 G06T 7/00 G06T 7/60 H01L 21/64- 21/66 H05K 13/00-13/04

Claims (8)

(57) [Claims] An imaging unit that divides the semiconductor device into predetermined pixels and takes an image to obtain a grayscale image; obtains binary image information based on the grayscale image obtained by the imaging unit; An image processing apparatus for performing positioning, comprising: a lead tip temporary position measuring means for measuring a temporary lead tip position for each lead of a semiconductor device based on the binary image information; True lead tip position calculating means for calculating a true lead tip position for each lead from the grayscale image of the area, and a center of gravity for calculating the center of gravity and inclination of the semiconductor device based on the true lead tip position calculated by the true lead tip position calculating means. An image processing apparatus comprising: a calculation unit. 2. The image processing apparatus according to claim 1, wherein the lead tip temporary position measuring means mounts the semiconductor device such that the arrangement direction of the leads is substantially coincident with the horizontal scanning direction. A predetermined pixel is moved in the vertical scanning direction perpendicular to the scanning direction and in the root direction of the lead, and an adjacent lead is detected by scanning in the horizontal scanning direction. A temporary lead end position of an adjacent lead is obtained, and the above-described scanning is repeated to sequentially obtain a temporary lead end position of an adjacent lead. 3. The image processing apparatus according to claim 1, wherein the difference between the number of temporary lead tip positions obtained by the temporary lead tip position measuring means and a predetermined number of semiconductor device leads is equal to or greater than a predetermined value. In the case of (1), an erroneous recognition detecting means for recognizing that the detected temporary lead tip position is incorrect is provided. 4. The image processing apparatus according to claim 1, wherein the lead tip provisional position measuring means performs scanning in a horizontal scanning direction substantially coincident with a lead arranging direction, and determines a predetermined number of pixels in a portion recognized as an object. When the value is equal to or less than the value, the image processing apparatus determines that the read is not a read, and when it is determined that the read is not a read, does not perform the calculation of the provisional position of the leading end of the lead and performs scanning in the horizontal scanning direction. 5. A semiconductor device is divided into predetermined pixels and imaged to obtain a grayscale image, and binarized image information is obtained based on the grayscale image obtained by the imaging means to position the semiconductor device. An image processing method to be performed, comprising: measuring a tentative lead tip position for each lead of a semiconductor device based on the binarized image information; (C) calculating a true lead tip position and calculating a center of gravity and an inclination of the semiconductor device based on the calculated true lead tip position. 6. The image processing method according to claim 5, wherein the semiconductor device is placed in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and a vertical scanning direction perpendicular to the horizontal scanning direction is provided from the temporary lead tip position. In addition, it moves by a predetermined pixel in the root direction of the lead, detects adjacent leads by scanning in the horizontal scanning direction, and then moves in the vertical scanning direction and the tip direction of the leads to detect adjacent leads. An image processing method comprising: obtaining a temporary lead tip position; and repeating the scanning to sequentially determine a temporary lead tip position of an adjacent lead. 7. The image processing method according to claim 5, wherein a difference between the number of temporary lead tip positions obtained by the lead tip temporary position measuring means and a predetermined number of semiconductor device leads is equal to or more than a predetermined value. In the case of, it is recognized that the detected temporary lead tip position is incorrect,
An image processing method characterized by the following. 8. The image processing method according to claim 5, wherein the lead tip provisional position measuring means performs scanning in a horizontal scanning direction substantially coincident with the arrangement direction of the leads, and determines a predetermined number of pixels of a portion recognized as an object. If the value is equal to or less than the value, the image processing method is characterized in that it is determined that the read is not a read, and when it is determined that the read is not a read, the scan is performed in the horizontal scanning direction without calculating the temporary position of the leading end of the lead.