JPH11214415A - Positioning method by image processing - Google Patents

Abstract

(57) [Summary] [Problem] When a pellet is positioned in a misaligned state due to erroneous detection and measurement error in positioning measurement by image processing, the bonding apparatus recognizes the pellet at an incorrect position, so that bonding failure is caused. Occurs and lowers the yield. SOLUTION: By performing image processing, "read coordinate data" S1 is performed on a positioning point represented by a rectangular vertex,
"Calculation of parallelism between any pair of opposing sides" S2,
Check whether the parallelism is within the reference value or not by S3,
"Calculation of orthogonality" S4, the orthogonality between the side whose parallelism was inspected by S5 and "Whether the orthogonality is within the reference value" S5 and the adjacent side are checked, and if the parallelism is outside the reference value, Performs a parallelism and orthogonality inspection on another pair of opposite sides in the same manner, and furthermore, when both the parallelism and the orthogonality are outside the reference value,
According to "calculation of parallelism and orthogonality of all sides" S7, S13, and S14, the parallelism and orthogonality of all sides are inspected, and a highly accurate side (positioning point) is obtained.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a positioning method based on image processing, and more particularly to a positioning method based on image processing including a step of checking the accuracy of coordinate data read by image processing.

[0002]

2. Description of the Related Art In a conventional positioning method using image processing, positioning is performed using data read by image processing as it is. For example, positioning of a pellet in a bonding device such as a CCD pellet has been performed based on the assumption that data read by image processing is accurate.

[0003]

However, when the measurement unit for positioning by the image processing is erroneously detected or when the measurement by the image processing includes an error, the positioning is performed in a shifted state. For example, in the case of a bonding apparatus, the bonding position is shifted, so that the bonding accuracy is reduced.
As shown in FIG. 3A, accurate position data (p1 to
Although p4) should be sent from the image processing apparatus, as shown in (2) in FIG.
When the data of the erroneous detection position of 2, p3, p4 'is sent, that is, the inclination θ' of the line Q 'passing through p1' p4 'is the correct p1p4 shown in FIG. Even if the inclination is clearly different from the inclination θ of the passing line Q, the bonding apparatus will recognize the pellet at an incorrect position. As described above, if the bonding is performed without accurately recognizing the position of the pellet, a bonding failure occurs and the yield is reduced.

[0004]

SUMMARY OF THE INVENTION The present invention is a positioning method based on image processing which has been made to solve the above problems. First, coordinate data of a positioning point represented by a rectangular vertex is read by image processing. The parallelism of an arbitrary opposing side of the rectangle constituted by the coordinate data as a vertex is calculated, and the orthogonality between the side on which the parallelism is calculated and a side adjacent thereto is calculated. If the calculated parallelism is outside the reference value, the parallelism of another opposing side is calculated, and the orthogonality of the calculated side with the adjacent side is calculated. When the parallelism calculated in this way and at least one of the orthogonalities calculated above are within the reference value, the side to be determined is at least one of the sides on which the parallelism is calculated. On the other hand, when both the calculated parallelism and orthogonality are outside the reference values, the parallelism and orthogonality are inspected for all sides of the rectangle formed based on the coordinate data.

The inspection of the parallelism and orthogonality of all sides of the rectangle is performed by calculating the parallelism of a pair of opposite sides of the rectangle and the parallelism of another pair of opposite sides of the rectangle. By calculating all orthogonalities between adjacent sides of the rectangle, selecting a combination of sides having the highest calculated parallelism or orthogonality, and calculating an average of the selected two sides.

In the positioning method based on the image processing, the parallelism of opposing sides of the rectangle and the orthogonality of adjacent sides are calculated, and the side having the highest parallelism or orthogonality is selected. That is, in a rectangle (rectangle and square), two opposing sides are parallel and all adjacent two sides form a right angle. Therefore, the opposing two sides of the rectangle formed based on the coordinate data read by the image processing are used as a reference. By calculating the degree of orthogonality with a side adjacent to one of the sides on which the degree of parallelism has been calculated and the degree of parallelism has been calculated, the calculated degree of parallelism and at least one of the obtained degrees of orthogonality are reference values. If the value is within the value, the side for which the parallelism is obtained is the side within the reference value that is shifted from the accurate position. Then, since the points at both ends of the side are coordinate data, a deviation from the correct position of the two points is a positioning point within the reference value.

If the parallelism calculated first is outside the reference value, the same calculation is performed on the other two opposite sides of the rectangle. Is found within the reference value. Then, since the points at both ends of the side are coordinate data, a deviation from the correct position of the two points is a positioning point within the reference value.

Further, when both the calculated parallelism and orthogonality are outside the reference values, the parallelism and orthogonality are calculated for all sides of the rectangle formed based on the coordinate data, and each calculation is performed. By selecting a combination side having a high degree of parallelism or orthogonality and calculating the average of the two sides, the side with the smallest deviation from the correct position is determined. Since the points at both ends of the side are the coordinate data, the two points are the positioning points where the deviation from the accurate position is minimized.

As a result, the coordinate data read with an error is obtained as accurate coordinate data or coordinate data closest to the accurate coordinate data.

[0010]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS A basic configuration of a positioning method using image processing according to the present invention will be described. In this positioning method, first, coordinate data of a positioning point represented by a rectangular vertex is read by image processing. The rectangle at that time is a rectangle or a square. Then, the parallelism of any opposing side of the rectangle formed based on the coordinate data is calculated, and the orthogonality between the side on which the parallelism is calculated and the side adjacent thereto is calculated.

If the calculated parallelism and at least one of the calculated orthogonalities are within the reference value, at least one of the sides on which the parallelism is calculated is determined as the side to be determined.

On the other hand, when the calculated parallelism is out of the reference value, the parallelism of another opposite side is calculated, and the orthogonality between the calculated side and the side adjacent thereto is calculated. . If the calculated degree of parallelism and at least one of the calculated degrees of orthogonality are within the reference value, at least one of the sides on which the calculated degree of parallelism is calculated is determined as the side to be determined.

On the other hand, when both the calculated parallelism and orthogonality are outside the reference values, the parallelism and orthogonality are calculated for all sides of the rectangle formed on the basis of the coordinate data. The calculated side having a high degree of parallelism or orthogonality is selected, the average of the two sides is obtained, and the average is determined as the side to be obtained.

The above positioning method is applied to the following cases as an example. For example, in the image processing of the pellet position performed at the time of performing bonding, a corner point of the pellet is sent to the bonding apparatus as coordinate data indicating the position. The bonding apparatus calculates the θ direction (rotation direction) of the bonding position based on the coordinate data sent first. In the calculation of the bonding position in the θ direction, θ can be calculated if two accurate points among the four points of the image processing data are known. The two points here are points that define the position of each side of the pellet when viewed in plan. Before the calculation in the θ direction, the accuracy of the coordinate data is checked. At that time, the positioning method described above is adopted.

Next, the details of the embodiment relating to the positioning method based on the image processing of the present invention will be described with reference to the flowchart of FIG. 1 and the explanatory view of the coordinate data and rectangular data of FIG. Here, as an example, a method of positioning a pellet during bonding will be described.

As shown in FIG. 1, "reading of coordinate data" S1 is performed by "beginning". That is, a step of reading coordinate data of a positioning point represented by a vertex of a rectangle by image processing is performed. For example, as shown in (1) of FIG. 2, the coordinate data of each of the points p1 to p4 constituting the positioning coordinates set on the pellet 11 at the time of bonding is read. For example, the points p1 to p4 are coordinates of the corner of the pellet. These p1 to p4 are located, for example, at the vertices of the rectangle 21, and the rectangle may be a rectangular pellet as shown or a square pellet (not shown). The rectangle 21 here is the outer shape of the pellet when viewed in plan.

Next, a "calculation of parallelism between any pair of opposing sides" S2 is a first calculation for calculating the parallelism between any pair of opposing sides of a rectangle formed based on the coordinate data. Perform the process. Here, as shown in FIG. 2B, opposing sides t0 (hereinafter, the side t0 is a line segment p1p2) and a side t2 (hereinafter, the side t2 is a line segment p3p4) of the rectangle 21 read by the image processing. ) Is calculated. The sides t0 and t1 (hereinafter, the side t1 is a line segment p2p3), t2, and t3 (the side t3 is a line segment p4p1) of the rectangle 21 indicate the respective sides of the pellet when viewed in plan. . The parallelism refers to an angle between two sides when the two sides are moved in parallel and crossed. The following parallelism is defined as above.

Subsequently, in "whether or not the parallelism is within a reference value" S3, a first determination step of determining whether or not the parallelism calculated in the first calculation step is within a reference value is performed. Here, for example, assuming that the reference value is 0.2 °, the determination is made based on whether the parallelism is 0.2 ° or less or greater than 0.2 °. In this manner, the inspection of the parallelism between any pair of opposing sides is performed.

In the parallelism determination in the first determination step, when the parallelism is "Yes" within the reference value,
In the "calculation of orthogonality" S4, a second calculation step of calculating the orthogonality between the side on which the parallelism is calculated and an adjacent side is performed. Here, the orthogonality refers to the absolute value of the difference between the angle at which the two sides for which the orthogonality is determined and the right angle. Less than,
The orthogonality is defined in this way. For example, each orthogonality between the side t0 and the side t1 adjacent thereto and the side t0 and the side t3 adjacent thereto is calculated.

Next, in "whether the orthogonality is within a reference value" S5, a second determination step of determining whether or not the orthogonality calculated in the second operation step is within a reference value is performed. Here, for example, assuming that the reference value is 0.2 °, the determination is made based on whether the orthogonality is 0.2 ° or less or greater than 0.2 °. In this manner, the inspection of the orthogonality between the side on which the parallelism is calculated and the side adjacent thereto is performed.

In the determination of the degree of orthogonality in the second determination step, if at least one of the obtained degrees of orthogonality is "Yes" within the reference value, the "determined side is the side on which the parallelism is calculated" S6 The side to be sought is determined, and the process ends. For example, if any of the orthogonality between the side t0 and the side t1 adjacent thereto and the orthogonality between the side t0 and the side t3 adjacent thereto are within the reference value, the side t0 or the side t0
The side t2 whose parallelism is within the reference value is the side to be obtained.
That is, the sides t0 and t2 are sides on which the parallelism is calculated in the first calculation step.

On the other hand, in the determination in the second determination step,
When the orthogonality is out of the reference value “No”, the “inspection of parallelism and orthogonality of all sides” S7 is performed, and the parallelism is applied to all sides of the rectangle formed based on the coordinate data. And inspection of orthogonality. The details of the “inspection of parallelism and orthogonality of all sides” will be described later in detail.

On the other hand, in the determination in the first determination step,
When the degree of parallelism is “No” outside the reference value, the “calculation of the degree of parallelism between another pair of opposing sides” S <b> 8 indicates that the degree of opposition other than the degree of parallelism used in the first degree of calculation in the first calculation step is different A third calculation step of calculating the parallelism between the pair of sides is performed.
Here, as shown in FIG. 2B, the parallelism between the opposing sides t1 and t3 of the rectangle 21 read by the image processing is calculated.

Subsequently, in "whether or not the parallelism is within a reference value" S9, a third determination step of determining whether or not the parallelism calculated in the third calculation step is within a reference value is performed. Here, for example, assuming that the reference value is 0.2 °, the determination is made based on whether the parallelism is 0.2 ° or less or greater than 0.2 °. In this way, the inspection of the parallelism between another pair of opposite sides is performed.

In the determination of the parallelism in the third determination step, if the parallelism is "Yes" within the reference value,
In "calculation of orthogonality" S10, a fourth calculation step of calculating the orthogonality between the side on which the parallelism is calculated and the side adjacent thereto is performed. For example, the side t1 and the sides t0 and t1 adjacent to the side t1
And each orthogonality between the edge and the adjacent side t2 is calculated.

Next, "whether the orthogonality is within the reference value" S11
Thus, a fourth determination step of determining whether the orthogonality calculated in the fourth calculation step is within the reference value is performed. Here, for example, assuming that the reference value is 0.2 °, the orthogonality is 0.2 ° or less,
Alternatively, it is determined based on a criterion of whether the angle is larger than 0.2 °. In this manner, the inspection of the orthogonality between the side on which the parallelism is calculated and the side adjacent thereto is performed.

In the determination of the degree of orthogonality in the fourth determination step, if at least one of the obtained degrees of orthogonality is “Yes” within the reference value, the “determined side is the side on which the parallelism is calculated” S12 Decide which side you want, and click "End"
Becomes For example, if either the orthogonality between the side t1 and the adjacent side t0 and the orthogonality between the side t1 and the adjacent side t2 are within the reference value, the side t1 or the side t
The side t3 whose parallelism with 1 is within the reference value is the side to be obtained. That is, the sides t1 and t3 are sides on which the parallelism is calculated in the third calculation step.

On the other hand, in the determination in the fourth determination step,
When the orthogonality is out of the reference value “No”, the “inspection of parallelism and orthogonality of all sides” S13 is performed to execute parallelism with respect to all sides of the rectangle configured based on the coordinate data. And inspection of orthogonality. The “inspection of parallelism and orthogonality of all sides” S13 is the “inspection of parallelism and orthogonality of all sides”.
It is the same as that of S7.

On the other hand, if the orthogonality in the determination in the third determination step is "No" outside the reference value, the "inspection of the parallelism and orthogonality of all sides" S14 is performed, so that the coordinate data is obtained. Inspection of parallelism and orthogonality is performed on all sides of a rectangle formed based on the above. This "inspection of parallelism and orthogonality of all sides"
S14 is the same as that of the above-mentioned "inspection of parallelism and orthogonality of all sides" S7.

Next, the above "inspection of parallelism and orthogonality of all sides"
Will be described. The parallelism of a pair of opposing sides of the rectangle and the parallelism of another pair of opposing sides of the rectangle are calculated. For example, the parallelism between the side t0 and the side t2 of the rectangle 21 as shown in FIG.
3 is calculated. At the same time, the orthogonality between each adjacent side of the rectangle 21 is calculated. For example, the orthogonality between the sides t0 and t1, the orthogonality between the sides t1 and t2, the orthogonality between the sides t2 and t3, and the orthogonality between the sides t3 and t0 are calculated. Then, a combination of sides having the highest calculated degree of parallelism or orthogonality is selected, and an average of the selected two sides is obtained.

In the positioning method based on the image processing described with reference to FIGS. 1 and 2, the parallelism of opposing sides of a rectangle and the orthogonality of adjacent sides are calculated, and the side having the highest parallelism or orthogonality is calculated. Is selected. A rectangle (rectangle and square) has two opposing sides parallel and all adjacent two
Since the sides form a right angle, two opposite sides of the rectangular data read by the image processing have a parallelism within the reference value, and are orthogonal to the side adjacent to one of the sides on which the parallelism is calculated. By checking the degree, that is, "calculation of parallelism between any pair of opposing sides" S2, "whether or not parallelism is within a reference value" S3, "calculation of orthogonality" S4, and "orthogonality" Whether the degree is within the reference value ”S5
If the calculated parallelism and at least one of the calculated orthogonalities are within the reference value, the side from which the parallelism is obtained is the side within the reference value that is shifted from the accurate position. Further, since the points at both ends of the side are coordinate data, two points, that is, positioning points whose deviation from an accurate position is within the reference value, are selected.

If the parallelism calculated first is outside the reference value, the parallelism and orthogonality are checked on another two opposite sides, that is, "another pair of opposite sides is checked. Of parallelism calculated by performing “calculation of parallelism” S8, “whether or not the parallelism is within a reference value” S9, “calculation of orthogonality” S10, and “whether or not the orthogonality is within a reference value” S11 If at least one of the calculated orthogonalities is within the reference value, the side from which the parallelism is obtained is the side within the reference value that deviates from the correct position. Further, since the points at both ends of the side are coordinate data, two points, that is, positioning points whose deviation from an accurate position is within the reference value, are selected.

Further, when both the calculated parallelism and orthogonality are outside the reference values, the parallelism and orthogonality are calculated for all sides of the rectangle formed based on the coordinate data, ie, By performing the “inspection of parallelism and orthogonality of all sides” S7, S13, and S14, a side of each combination having a high calculated parallelism or orthogonality is selected, and the average of the two sides is obtained. The side with the smallest deviation from the correct position is determined. In addition, since the points at both ends of the side are coordinate data, two points, that is, positioning points that minimize the deviation from an accurate position, are selected.

As a result, the coordinate data read with an error is obtained as accurate coordinate data or coordinate data closest to the accurate coordinate data.

The above-described positioning method using image processing can be used for positioning a pellet at the time of bonding. That is, before positioning using the coordinate data read by the image processing, the accuracy of the read coordinate data is checked by the positioning method described above. Thereby, positioning can be performed based on highly accurate coordinate data. By improving the positioning accuracy in this way, it is possible to improve the accuracy of the positioning of the pellets in a pellet bonding apparatus such as a CCD pellet.

[0036]

As described above, according to the present invention,
By inspecting the parallelism of two opposing sides of a rectangle formed based on the coordinate data read by the image processing, and inspecting the orthogonality with the side adjacent to one of the sides whose parallelism has been inspected. It is possible to determine the side where the deviation from the accurate position is within the reference value or the side where the deviation from the accurate position is the smallest, so that the two points with the highest accuracy are determined from both ends of the determined side. It is possible to find points. Therefore, for example, θ of the pellet at the time of bonding
When detecting the position in the direction, the θ direction can be detected from two highly accurate positioning points, so that the bonding accuracy can be improved and the yield in the bonding process can be improved.

[Brief description of the drawings]

FIG. 1 is a flowchart of an embodiment relating to a positioning method based on image processing of the present invention.

FIG. 2 is an explanatory diagram of coordinate data and rectangular data.

FIG. 3 is an explanatory diagram of a problem.

[Explanation of symbols]

S1 ... "reading of coordinate data", S2 ... "calculation of parallelism between any pair of opposing sides", S3, S9 ... "whether or not the parallelism is within a reference value", S4, S10 ... "orthogonality" , S5, S11... "Whether the orthogonality is within the reference value", S5
6, S12... "The side to be calculated is the side on which the parallelism is calculated", S
7, S13, S14 ... "calculation of parallelism and orthogonality of all sides", S8 ... "calculation of parallelism of another pair of opposite sides"

Claims (4)

[Claims] An image processing device reads coordinate data of a positioning point represented by a vertex of a rectangle, calculates parallelism of an arbitrary opposing side of a rectangle having the coordinate data as a vertex, and calculates the parallelism of the parallel data. Calculating the degree of orthogonality between the side on which the degree was calculated and the side adjacent thereto, and calculating the degree of parallelism of another opposite side if the calculated degree of parallelism is outside the reference value, and calculating the degree of parallelism Calculating the degree of orthogonality with the side adjacent to the calculated side, and when the calculated degree of parallelism and at least one of the calculated degrees of orthogonality are within a reference value, at least one of the sides on which the calculated side is calculated. When both the calculated parallelism and orthogonality are outside the reference values, the inspection of the parallelism and orthogonality is performed on all sides of the rectangle formed based on the coordinate data. Rank by image processing How to decide. 2. The positioning method according to claim 1, wherein the coordinate data of a positioning point represented by a vertex of the rectangle is read by the image processing; A first calculation step of calculating the parallelism between any pair of opposing sides, a first determination step of determining whether the parallelism calculated in the first calculation step is within a reference value, and the parallelism. In the determination of the above, if the parallelism is within the reference value, a second calculation step of calculating the orthogonality between the side on which the parallelism is calculated and an adjacent side, and the orthogonality calculated in the second calculation step is A second determination step of determining whether or not the value is within a reference value; and if at least one of the orthogonalities in the determination of the second determination step is within the reference value, the side to be determined is the side on which the parallelism is calculated. And the determination in the second determination step When the orthogonality in the rectangular shape is outside the reference value, a step of inspecting the parallelism and the orthogonality for all sides of the rectangle formed based on the coordinate data; and the parallelism in the determination in the first determination step. Is outside the reference value, a third calculation step of calculating the parallelism between a pair of opposing sides other than the calculation of the parallelism, and whether the parallelism calculated in the third calculation step is within the reference value. A third determining step of determining whether or not the parallelism in the determination in the third determining step is within a reference value; and a fourth calculating step of calculating an orthogonality between the side on which the parallelism is calculated and an adjacent side. A calculating step, a fourth determining step of determining whether or not the orthogonality calculated in the fourth calculating step is within a reference value, and at least one of the orthogonalities in the determination of the fourth determining step is within a reference value. , The side to be determined is the side on which the parallelism is calculated in the third calculation step. And if the orthogonality in the determination of the fourth determination step is outside the reference value, a step of inspecting the parallelism and orthogonality for all sides of the rectangle configured based on the coordinate data, When the orthogonality in the determination in the third determination step is outside the reference value, a step of inspecting the parallelism and the orthogonality for all sides of the rectangle configured based on the coordinate data. A positioning method by image processing characterized by the following. 3. The positioning method according to claim 1, wherein the inspection of the parallelism and the orthogonality with respect to all sides of the rectangle based on the coordinate data is performed by checking the parallelism of a pair of opposite sides in the rectangle. And the degree of parallelism of the other pair of opposing sides of the rectangle are calculated, and the degree of orthogonality between adjacent sides of the rectangle is calculated. And determining the average of the two sides of the image. 4. The positioning method according to claim 2, wherein the inspection of all sides of the rectangle configured based on the coordinate data includes the parallelism of a pair of opposite sides of the rectangle and the parallelism of the rectangle. Along with calculating the parallelism of the other pair of opposing sides, the orthogonality of each adjacent side of the rectangle is calculated, and the side of the combination having the highest calculated parallelism or orthogonality is selected. A positioning method based on image processing, wherein an average of two sides is obtained.