JP2003256808A - Image processor and image processing method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image processor resistant to a noise on an arbitrarily shaped electronic component or an image signal by efficiently recognizing the electronic component having luminance changing points other than a lead even when luminance changing points exist in a background in the surrounding of the lead part of the electronic component to be recognized by a common algorithm. <P>SOLUTION: The position information of a plurality of model feature points(luminance changing points) related with an objective lead group 30a to be outputted by a model shape information storing means 8 are compared with the position information of a plurality of feature point candidates 31-42 of lead groups 30a and 30b obtained from the image signal, and the combination of the feature point candidates which are the closest to the relative position relation of the model feature points(luminance feature points) is selected, and the position of the objective lead group 30a is calculated from the position information of the combination of the selected feature point candidates. <P>COPYRIGHT: (C)2003,JPO

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 and a processing method necessary for correcting the position of an electronic component such as an industrial robot or a component mounter for mounting an electronic component on a printed circuit board.

[0002]

2. Description of the Related Art Conventionally, an image pickup device such as a CCD camera has been used to correct the position of a component such as an industrial robot or a component mounter for mounting an electronic component (hereinafter, also referred to as "component") on a printed circuit board. The position of a component is detected by visually recognizing the component, and the position and angle of the component are corrected based on the detected component position information. In such image recognition of an electronic component, for example, an edge checker or the like is provided which scans an image signal obtained from the target electronic component to detect a brightness change point, and the shape of the electronic component to be detected. The position of this edge checker is decided according to.

There are two methods for detecting the position and angle of a component by this image recognition. One is a method for calculating the position and angle of the component by detecting the electrodes of the component, and the other is the outer shape of the component. It is a method of obtaining the position and angle of the component from the.

A method of detecting the electrode of a component is, for example, disclosed in Japanese Patent No. 3225067, by using a camera to detect a luminance signal of an electrode (hereinafter, also referred to as "lead") on a scanning line of an image signal. It describes a method of detecting a light-dark change point from a signal obtained by differentiating to obtain an edge center position of the leads and calculating a distance or an angle between the leads.

The patent No. 3 is shown in FIGS. 11 (a), 11 (b) and 11 (c).
A method for detecting electrodes of the component described in Japanese Patent No. 225067 will be described. FIG. 11A shows how a luminance signal is detected by the horizontal scanning line signal SX on the image signal for, for example, three leads 90 having a flaw 90d. 90L and 90R represent the left and right edges of the lead 90, respectively.

FIG. 11B shows luminance signals B1, B2, B for each lead 90 obtained by the horizontal scanning line signal SX.
3 is shown. N is 90 scratches on each lead
represents the noise signal caused by d.

FIG. 11C shows each of the luminance signals B1, B2.
The differential signal obtained by differentiating B3 is shown. At this time, the noise differential signal NC by the noise signal N is also detected.

In order to eliminate the influence of the noise differential signal NC due to the noise signal N in identifying the edge position of each lead, as shown in FIG. 11C, for example, when the differential signal is detected, the left end in the figure is detected. Actual left edge 9 of lead 90
Outside position D on the left of the position where 0L is predicted to exist
The differential signal of the luminance signal B1 is detected from L1 to the right, and the differential signal XL1 detected first is recognized as the differential signal by the left edge 90L.

Similarly, a process of detecting a differential signal of the luminance signal B1 from the outer position DR1 on the right side of the position where the actual right edge 90R of a certain lead 90 is predicted to exist to the left side is first detected. The differentiated signal XR1 is recognized as the differentiated signal by the right edge 90R.

In this manner, the left and right edge positions of the lead 90 are specified and the center position of the lead 90 is specified regardless of the noise differential signal NC due to the noise signal N formed at the center of the differential signal of the lead 90 (between XL1 and XR1). Can be calculated. The above-described differential signal detection processing is performed for each lead, the differential signals XL2 and XR2 of the lead of each lead, for example, in the center of the drawing are detected, the edge position is specified, and the center position is calculated.

By using the center position of each lead, for example, the center position of the leads or the center position of the whole lead can be obtained.

[0012]

However, in the image recognition of the lead portion of the electronic component, a detection line such as a scanning line is placed at a position where the target lead is supposed to exist in the image signal of the electronic component. The target lead portion position is calculated by assuming that all the brightness change points detected on the detection straight line are the brightness change points due to the read.

Therefore, it is necessary to pay attention to the adjustment of the illumination and the arrangement of the background so that the brightness change points other than the lead do not occur on the detection straight line in the image signal.

Further, in the case of a component which may have a brightness change point other than the lead to be detected, it is necessary to set some assumptions or conditions and select only the brightness change point by the target lead. In this case, for example, when the shape of the part changes, the assumptions and conditions of the lead detection line must be changed, and an image processing algorithm corresponding to the shape of the part is required each time.

In view of the above point, the present invention efficiently recognizes an electronic component having a luminance change point other than the lead and a luminance change point in the background near the lead portion of the electronic component to be recognized with a common algorithm. It is an object of the present invention to propose an image processing apparatus and a processing method that are resistant to noises on electronic components and image signals of arbitrary shapes by performing the above.

[0016]

The image processing apparatus of the present invention comprises:
An image pickup means for picking up an image of a lead group of an electronic component, a model shape information storage means for storing model shape information of a plurality of lead groups of the electronic component, and an electronic component having a plurality of lead groups obtained by the image pickup means. In an image processing apparatus having an image processing means for processing an image signal, the image processing means obtains from the position information of a plurality of model feature points regarding the target lead group output from the model shape information storage means and the image signal. The position information of the combination of the selected feature point candidates is selected by comparing the position information of the plurality of feature point candidates of the selected lead group, and the combination of the feature point candidates that is closest to the relative positional relationship of this model feature point is selected. From this, the position of this target lead group is calculated. Here, the position information of the model feature point of this electronic component is calculated from the position coordinates of the brightness change point when the detection straight line is set in the predetermined direction of the target lead group, and the position information of this feature point candidate is, for example, The detection straight line is applied to this image signal to calculate from the position coordinates when the brightness change point of the target lead group is detected.

According to the present invention, the position information of the feature point candidate of the target lead group detected from the image signal is compared with the position information of the model feature point of the target lead group output from the model shape information storage device. Then, the combination of the feature point candidates closest to the model feature point is selected, and the target lead group position in the image signal is calculated from the position information of the selected combination of the feature point candidates. The target lead group position can be calculated by removing the influence of unnecessary feature point candidates, and image processing can be performed that is resistant to irregularly shaped lead groups and noise on the image signal.

Further, the image processing apparatus of the present invention selects, from the feature point candidates, a combination of two points which are estimated to correspond to the two points having the largest distance among the plurality of model feature points. The image processing is performed by comparing the positions of the model feature points and the feature point candidates only for combinations of feature point candidates including points.

According to the present invention, a plurality of combinations of two points which are considered to correspond to the two points having the largest distance among the plurality of model characteristic points are selected from the characteristic point candidates, and the two points are selected.
Since the positions of the model feature points and the feature point candidates are compared only for combinations including points, the calculation time can be shortened by avoiding brute force calculation in all cases.

Further, the image processing apparatus of the present invention, when comparing the position of this model feature point and this feature point candidate, the brightness change point near the position estimated from the position information of the model feature point on the detection line of the image signal. A combination of feature point candidates that does not exist is determined to be an incorrect combination and the subsequent comparison calculation is not performed.

According to the present invention, regarding the combination of two points selected from the feature point candidates, the combination in which the brightness change point does not exist near the position estimated from the position information of the model feature point is a wrong combination. Since it is determined that the comparison calculation is not performed thereafter, it is possible to avoid unnecessary calculation and perform calculation efficiently.

The image processing method of the present invention is obtained by the image pickup means for picking up an image of a lead group of an electronic component, the model shape information storage means for storing model shape information of a plurality of lead groups of the electronic component, and the image pickup means. And image processing means for processing an image signal of an electronic component having a plurality of lead groups, and the image processing means outputs the positions of the plurality of model feature points with respect to the target lead group output from the model shape information storage means. The information is compared with the position information of a plurality of feature point candidates of the lead group obtained from this image signal, the relative positional relationship of this model feature point is selected, and the combination of this feature point candidate closest to the selected feature is selected. The position of this target lead group is calculated from the position information of the combination of point candidates. Here, the position information of the model feature point of this electronic component is calculated from the position coordinates of the brightness change point when the detection straight line is set in the predetermined direction of the target lead group, and the position information of this feature point candidate is, for example, The detection straight line is applied to this image signal to calculate from the position coordinates when the brightness change point of the target lead group is detected.

Further, the image processing method of the present invention selects from the feature point candidates a combination of two points which are estimated to correspond to the two points having the largest distance among the plurality of model feature points. The image processing is performed by comparing the positions of the model feature points and the feature point candidates only for combinations of feature point candidates including points.

Further, according to the image processing method of the present invention, at the time of comparing the positions of the model feature points and the feature point candidates, the brightness change point near the position estimated from the position information of the model feature points on the detection straight line of the image signal. A combination of feature point candidates that does not exist is determined to be an incorrect combination and the subsequent comparison calculation is not performed.

[0025]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of an image processing apparatus and a processing method according to the present invention will be described below with reference to the drawings.

FIG. 1 is a schematic configuration diagram showing an example of an electronic component mounter to which the image processing apparatus of the present invention is applied. This electronic component mounter is an electronic component (hereinafter,
A holding unit 2 for holding a "component") 1, and this holding unit 2
A driving device 3 for driving the device 1, an oblique reflection illumination 4 for irradiating the component 1 with light from an oblique position, a video camera 5 as an image pickup means for imaging the component 1, and an image signal from the video camera 5 temporarily. An image signal processing / control unit 6 for processing and controlling an image signal including an image memory unit 7a for storing, a calculation unit 7b, a model shape information storage device 8 and the like, and an image signal processing result in the image signal processing / control unit 6. Is composed of a communication means 9 for transmitting to the drive device 3.

In FIG. 1 configured as described above, the holding unit 2 driven by the drive unit 3 sucks and holds the component 1 by chuck means such as vacuum suction, and holds and fixes the component 1 at a predetermined position in the air, for example. . The oblique reflection illumination 4 irradiates the component 1 held in the air with light from, for example, an obliquely lower position, and the coaxial reflection illumination 5b also irradiates the light via the prism 5a. At this time, in order to image the background of the component 1 black, a black plate 4a or the like is installed on the back surface of the component 1 on the side opposite to the oblique reflection illumination 4 and the coaxial reflection illumination 5b.

The image of the component 1 irradiated with light is the prism 5a.
After that, the image is captured by the video camera 5 and captured.
In the video camera 5, the image of the component 1 is converted into an image signal,
It is sent to the image signal processing / control unit 6.

The image signal processing / control section 6 processes the image signal from the video camera 5 and controls the drive unit 3 based on the processing result. The image signal obtained from the video camera 5 is temporarily stored in the image memory unit 7a and transmitted to the calculation unit 7b, and stored in the component shape on the image signal and the model shape information storage device 8 by the calculation unit 7b. The existing model shape information is compared.

The model shape information storage device 8 is composed of, for example, a ROM (Read Only Memory) or the like, model shape information and center positions of lead groups of a plurality of parts,
And information and the like obtained in the process of calculating the center position.

Here, as a result of the above-mentioned comparison, for example, the position of the component 1 on the image signal is greatly bent, or due to a lead defect or the like, the component 1 and the model of the component registered in the model shape information storage device 8 are obtained. When it is determined that the shape information is different from the shape information, the image processing step is not performed because the line process is rejected as a difference in the type of parts or a defective product.

When it is determined that the part 1 matches the model shape information, image processing described later is performed based on the model shape information, and the center position and angle of the part 1 are calculated.
Then, the calculated component center position and angle information is transmitted to the drive device 3 via the communication means 9.

In this drive unit 3, X based on the component center position and angle information supplied from the image signal processing / control unit 6 is used.
A position correction operation is performed in the three-dimensional directions of the YZ axes and the rotation direction r to set the component 1 to a predetermined position on, for example, a printed circuit board, and then the electrodes of the component 1 are subjected to soldering processing and the like. It is made to be mounted on a printed circuit board.

Next, a method of detecting parts will be described.

FIG. 2 schematically shows image signals of parts obtained on a monitor (not shown) connected to the video camera 5, for example. In the figure, reference numeral 11 denotes a screen of a monitor or the like, and the background 12 imaged in black by the black plate 4a is indicated by diagonal lines. Reference numeral 10 shows an image of the imaged component 1, and a lead group 10a consisting of three long leads and a lead group 10b consisting of two short leads joined to one side of the component 10 are an oblique reflection illumination 4 and a coaxial reflection. The light from the illumination 5b is reflected and appears white. However, as shown by the diagonal lines in the figure, the black portion of the surface of the component, such as the black mold body 10c, is also imaged black.

When the reflection illumination such as the oblique reflection illumination 4 is used as in the present example, a black plate 4a is installed on the back side of the component 1 which irradiates light, and the background of the image of the component 1 is imaged black. To do so.

On the other hand, a diffuser plate is used in place of the black plate 4a, and a transmissive illumination is installed above the component 1 so that the coaxial reflective illumination 5b is not irradiated and the component 1 is irradiated with light from the transmissive illumination through the diffuser plate. Then, the diffuser plate diffuses the light and becomes white as a whole, so that the background color is reversed from that of the example in FIG. 2 and the background is imaged white, and the transmitted illumination creates a shadow on the component 1 and the lead portion and both are imaged black. .

Reflective illumination is used in this example,
Even when using transillumination, whether to detect a brightness change point, such as an edge checker, to detect a brightness change point from black to white or to detect a brightness change point from white to black. However, the common processing method can be used to calculate the component center position and angle based on the brightness change point.

FIG. 3 is a diagram showing an example of the one-dimensional luminance signal on the detection straight line.
0 is a black background 12 of a screen 11 such as a monitor (not shown)
Imaged above. In the detection straight line 14 which is set in the direction orthogonal to the lead for detecting the change in the brightness of the lead, the one-dimensional brightness signal 15 is detected along the scanning direction of the detection straight line 14 (the arrow from the right to the left in the drawing). Then, a differential signal 16 is generated by differentiating the one-dimensional luminance signal 15. And
Background 12 indicated by the above-mentioned 5 long and short leads and white circles
The one-dimensional luminance signal from the noise 13 above is detected.

FIG. 4 shows an example of a one-dimensional luminance signal when the detection straight line has a width. As in the example of FIG. 3, the white portion due to the reflection of the lead group 10a is imaged on the black background 12 of the screen 11, and the detection straight line 17 having a width for detecting the brightness change is the scanning direction of the detection straight line 17 (in the figure, The one-dimensional luminance signal 18 is detected along the direction from right to left), and then the differential signal 19 is generated by differentiating the one-dimensional luminance signal 18. Then, the one-dimensional luminance signal 1 from the above-mentioned long and short 5 leads and the noise 13 on the background 12 indicated by white circles
8 is detected.

In the case of the example of FIG. 4, since the detection straight line 17 is formed in a band shape having a certain width, the luminance is added in the width direction to detect the one-dimensional luminance signal 18, so that the noise 1
3 and the lead group 10b, which are not white in the width direction of the detection straight line, have less influence on the luminance signal, and can be easily distinguished from the detection target lead group having long leads, for example.

FIG. 5 shows an example of the luminance change point obtained from the differential signal 16 obtained by differentiating the one-dimensional luminance signal 15 of FIG. In the differentiated signal 16, the luminance change point at which the black changes from white to the white along the scanning direction from the right to the left has a positive peak, and the luminance change point at which the white changes to black has a negative peak. Here, the lead groups 10a and 1 are used as the brightness change points.
A total of 12 luminance change points, namely, 10 luminance change points obtained from the left and right edges of the 5 long and short leads constituting 0b and 2 places obtained from the noise 13 indicated by a white circle are detected.

The above-described image signal of FIG. 5 contains two types of lead groups, and for example, there is a brightness change point on the detection straight line due to the lead group 10b different from the lead group 10a to be detected. Also, in the figure, a white portion 13 due to the background exists on the left side of these lead groups. In such an image signal, when the peak of the differential signal is taken on the detection straight line in the figure to detect the brightness change point, as shown in FIG. 5, there is a brightness change point other than the brightness change point caused by the target lead group. 1-dimensional luminance signal. For such a signal, it is necessary to select one corresponding to the target lead from a plurality of brightness change points.

In the present example, one corresponding to the target lead group is selected from such a plurality of brightness change points and the position on the detection straight line, that is, the position on the one-dimensional coordinate is calculated.

Hereinafter, the center position of the lead group of the component is determined from the brightness change point of the detection straight line arranged at an arbitrary position of the component according to the flowchart showing the procedure of the image processing method of FIG. 6 and FIGS. The calculation algorithm will be described.

First, the position information of the model feature points is created from the shape information of the lead group of the component (step 1).

FIG. 7 schematically shows model shape information regarding, for example, a lead group of parts. In the figure, reference numeral 11 denotes a screen of a monitor or the like, and the background 12 imaged in black by the black plate 4a is indicated by diagonal lines. Reference numeral 10 denotes an image of the imaged component, which is a lead group 10a composed of three long leads and two short leads which are joined to one side of the component 10 and constitute five leads in which long and short leads are alternately arranged. The lead group 10b composed of leads looks white due to the light from the oblique reflection illumination 4 and the coaxial reflection illumination 5b being reflected.

In FIG. 7, the detection line 14 is set in advance on the screen 11 in a substantially horizontal direction from the right to the left of the component 10 by the model shape information storage device 8 in advance so as to detect the brightness change point. The position and the like of the detection straight line 14 are appropriately set by the model shape information storage device 8 in accordance with the lead group shape information of the component.

With respect to the lead group shape information to be detected, the first brightness change point of the brightness change points obtained when the above-described predetermined detection straight line is scanned is set as the start brightness change point. From the change point, the brightness change point distance for each brightness change point is calculated.

This start luminance change point serves as a reference when calculating the lead group center position described later. For example, in FIG. 7, a lead group 1 including three long leads
For 0a, the brightness change point 21 of the rightmost lead on the detection line 14 is set as the start brightness change point candidate, and the other 5
The distances x1, x2, x3, x4, x5 to the two brightness change points 22, 23, 24, 25, 26 are calculated as the brightness change point distances. In particular, the distance x5 from the start luminance change point 21 to the farthest luminance change point 26 is the maximum luminance change point distance 27.
And

Next, the brightness change point is calculated with respect to the detection straight line set in the image signal obtained from the components actually mounted (step 2).

FIG. 8 shows an example of image signals and brightness change points of actually mounted parts obtained on a monitor (not shown) connected to the video camera 5, for example. Reference numeral 30 in the drawing denotes a component, which is in a state of being slightly rotated from the regular mounting position on the screen 11. Screen 11 in the figure
A lead group 10a composed of three long leads and two short leads that constitute five leads in which long and short leads are alternately arranged by being joined to one side of the component 30 by the lateral detection straight line 14. The brightness change point between the lead group 10b and the white portion 43 on the background 12 and the background 12 is detected.

As a result, in the example of FIG. 8, the brightness change points 31, 32, 33, 34 between the five leads and the background 12 are shown.
35, 36, 37, 38, 39, 40, and 10 points on the left side of the 5 lead group, and a total of two points 41 and 42 on the left side of the non-lead white circle portion 43 on the background 12 and the brightness change points 41 and 42. Twelve brightness change points are sequentially detected on the detection line 14 from right to left in the arrow direction.
The coordinates on the one-dimensional coordinates of the above-mentioned 12 brightness change points are respectively X1, X2, X3, X4, X5, X6, X7,
Let them be X8, X9, X10, X11, and X12. In the following description, the brightness change point is also referred to as a feature point candidate of a component.

Next, the start luminance change point is obtained (step 3).

More specifically, the one having a brightness change point within a certain range, eg, ± 10%, which is extended from each brightness change point by the maximum brightness change point distance 27, is selected. The selected brightness change point is set as a start brightness change point candidate. Figure 8
In the example, it is determined that the brightness change points 40 and 42 exist as the brightness change points having the same polarity as that of the model feature point 26 near the positions separated by the maximum brightness change point distance 27 from the brightness change points 31 and 33, respectively. The brightness change points 31 and 33 are candidates for the start brightness change point.

Next, a squared error is obtained for each candidate start luminance change point (step 4).

First, the corrected luminance change point distance of each model feature point in which the luminance change point distance is corrected according to the maximum luminance change point distance is calculated. FIG. 9 shows an example in which the feature point candidate 31 is the start luminance change point candidate and the distance from the feature point candidate 31 to the feature point candidate 40 is the maximum luminance change point distance 27a.

Among the model feature points of the component 30, the brightness change point distance from the brightness change point 31 which is a candidate start brightness change point to the model feature points 22 and 23 is corrected according to the above-mentioned maximum brightness change point distance 27a. When the corrected luminance change point distance is calculated, correction model feature point candidates 22a and 23a are respectively set, and the coordinates on the one-dimensional coordinates at this time are set to x1 'and x2', respectively. The coordinate x1 ′ on the one-dimensional coordinate of the correction model feature point candidate 22a is x1 ′ = x1 / x5 × (X10−X1), and the coordinate x on the one-dimensional coordinate of the correction model feature point candidate 23a.
2'is calculated as x2 '= x2 / x5 * (X10-X1).

First, this correction model feature point candidate 2
It is searched whether there is a brightness change point near the coordinate x1 'of 2a.
In the case of FIG. 9, the feature point candidate 32 (coordinate X2) is the closest to the brightness change point of the same polarity, so the squared error between the correction model feature point candidate 22a and the brightness change point 32 is calculated, and this is started as described above. Feature point candidate 3 which is a brightness change point candidate
Add to the squared error about 1.

Next, similarly, the correction model feature point candidate 2
It is searched whether there is a luminance change point near the coordinate x2 'of 3a.
In the case of FIG. 9, the brightness change point 35 (coordinate X5) is the closest brightness change point of the same polarity. As described above, when there is a brightness change point near the correction model feature point candidate 23a, a square error between the brightness change point 35 and the correction model feature point candidate 22a is calculated and the start brightness change point candidate is calculated. It is added to the squared error of the feature point candidate 33.

The above-described calculation is performed for all the model feature points for the start luminance change point candidates to calculate the final squared error for the start luminance change point candidates.

Here, the case where the calculation of the squared error calculation for the above-mentioned start luminance change point candidate is terminated halfway will be described with reference to FIG. This is an example in which the feature point candidate 33 is a start luminance change point candidate and the distance from the feature point candidate 33 to the feature point candidate 42 is the maximum luminance change point distance 27b.

Among the model feature points of the component 30 described above, the distance from the brightness change point 33, which is a candidate start brightness change point, to the model feature points 22 and 23 is the brightness change point distance.
When the corrected brightness change point distances corrected according to the above-mentioned maximum brightness change point distance 27b are calculated, they are set as correction model feature point candidates 22b and 23b, respectively, and the coordinates on the one-dimensional coordinates at this time are x1 correction model feature points, respectively. Point candidate 22b
The coordinate x1 ″ on the one-dimensional coordinate of is x1 ″ = x1 / x5 × (X12−X3), and the coordinate x on the one-dimensional coordinate of the correction model feature point candidate 23b
2 ″ is calculated as x2 ″ = x2 / x5 × (X12−X3).

First, it is searched whether there is a brightness change point near the coordinate x1 "of the correction model feature point candidate 22b.
In the case of, the feature point candidate 34 (coordinate X4) is the closest as a brightness change point of the same polarity, but the coordinate X4 in the figure is ± 1 of the distance from the feature point candidate 33 (coordinate X3) to the coordinate x1 ″.
It is outside the range of 0%. That is, | (X4-X3) | <|
0.9 × x1 ″ |

As a result, the calculation unit 7b determines that the brightness change point in the vicinity of the correction model feature point candidate 22b does not exist and the feature point candidate 33, which is the above-mentioned start brightness change point candidate, is not the start brightness change point. Is not calculated.

As described above, the squared error calculation is performed for each of the start luminance change point candidates, and the squared error calculation is performed for all the model feature points without stopping the calculation halfway. Then, the one having the smallest square error is determined as the start luminance change point of the component 30 of the image signal (step 5).

Feature point candidates closest to each model feature point for this start luminance change point are set as feature points on the image signal, and the feature points of each lead are averaged from this feature point on the coordinates, for example, of the lead group. The lead group position is calculated by, for example, obtaining the center position (step 6).

Then, the component 3 calculated in this way
It is also possible to obtain positional information such as the relative center position of the component 30 from the positional information of the lead group 30a of 0.
The position information of 0 is, for example, the arithmetic unit 7 of the component mounter shown in FIG.
If the position correction of the electronic component 1 is performed by transmitting the electronic component 1 from b to the driving device 3 through the communication means 9, the electronic component 1 can be accurately mounted at a predetermined position such as a printed circuit board.

According to this example, the position information of the feature point candidates of the target lead group detected from the image signal is compared with the position information of the model feature point of the target lead group output from the model shape information storage device. Then, the combination of feature point candidates closest to the model feature point is selected, and the target lead group position in the image signal is calculated from the position information of the selected combination of feature point candidates. Even if the brightness change point is mixed with other lead groups or other brightness change points than the lead, or if there is a brightness change point in the background of the lead group to be recognized, the common algorithm effectively It is possible to calculate the target lead group position by removing the influence of unnecessary feature point candidates, and perform image processing that is resistant to irregularly shaped parts and noise on the image signal. Kill. This eliminates the need to adjust the illumination so that no brightness change points other than the leads or to make sure that there is no brightness change point in the background of the electronic component to be recognized, and various electronic components can be easily recognized. it can.

Further, according to this example, a plurality of combinations of two points which are considered to correspond to the two points having the largest distance among the plurality of model characteristic points are selected from the characteristic point candidates, and the combination including these two points is selected. Only in this case, the positions of the model feature points and the feature point candidates are compared, so that it is possible to avoid the brute force calculation in all cases and reduce the calculation time.

Further, according to this example, regarding a combination of two points selected from the feature point candidates, a combination having no brightness change point near the position estimated from the position information of the model feature point is an incorrect combination. Since the determination is made and the subsequent comparison calculation is not performed, useless calculation can be avoided and calculation can be performed efficiently.

The present invention is not limited to the above-mentioned examples, and it goes without saying that various other configurations can be adopted without departing from the gist of the present invention.

[0073]

According to the present invention, even a component in which a luminance change point due to a lead group to be detected and another lead group or a luminance change point other than the leads are mixed,
Also, even if there is a brightness change point in the background of the lead group to be recognized, the target lead group position can be effectively calculated by removing the influence of unnecessary feature point candidates by a common algorithm, and parts with irregular shapes can be calculated. There is an advantage that image processing that is resistant to noise on the image signal and the image signal can be performed. This allows
It is not necessary to adjust the illumination such that a brightness change point other than the lead does not occur, or to pay attention so that a brightness change point does not occur in the background of the electronic component to be recognized, and various electronic parts can be easily recognized.

Further, according to the present invention, a plurality of combinations of two points which are considered to correspond to the two points having the largest distance among the plurality of model characteristic points are selected from the characteristic point candidates, and the combination including these two points is selected. Only when the positions of the model feature points and the feature point candidates are compared, there is an advantage that the calculation time can be shortened by avoiding the brute force calculation in all cases.

Further, according to the present invention, regarding a combination of two points selected from the feature point candidates, a combination having no brightness change point near the position estimated from the position information of the model feature point is an incorrect combination. When the judgment is made and the subsequent comparison calculation is not performed, there is an advantage that the calculation can be avoided efficiently and uselessly.

[Brief description of drawings]

FIG. 1 is a schematic configuration diagram showing an example of an embodiment of an image processing apparatus of the present invention.

FIG. 2 is a diagram showing an example of an image signal obtained from an electronic component.

FIG. 3 is a diagram showing an example of a one-dimensional luminance signal on a detection straight line.

FIG. 4 is a diagram showing an example of a one-dimensional luminance signal when a detection straight line has a width.

FIG. 5 is a diagram showing an example of brightness change points.

FIG. 6 is a flowchart showing the procedure of the image processing method of the present invention.

FIG. 7 is a diagram for explaining the creation of position information of model feature points.

FIG. 8 is a diagram showing an example of an image signal actually captured and a luminance change point.

FIG. 9 is a diagram for explaining calculation of start luminance change point candidates.

FIG. 10 is a diagram for explaining calculation of start luminance change point candidates.

FIG. 11 is a diagram for explaining a conventional image processing method.

[Explanation of symbols]

1, 10, 30, ... Electronic parts, 5 ... Video camera, 6
.... Image signal processing / control unit, 7a ... Image memory unit, 7
b ... Calculation unit, 8 ... Model shape information storage device, 10
a, 10b, 30a, 30b ... Lead group, 14, 17
.... Detection line, 20, 21, 22, 23, 24, 25,
26, 31, 32, 33, 34, 35, 36, 37, 3
8, 39, 40, 41, 42, ... Luminance change point, 27, 2
7a, 27b ... ・ Maximum brightness change point distance

Continued front page    F term (reference) 2F065 AA01 CC27 FF04 HH12 HH13                       JJ03 LL46 PP01 QQ24 QQ31                       RR05                 5B057 AA03 BA02 DA07 DB02 DB05                       DB09 DC06 DC08 DC09 DC16                       DC33                 5E313 AA04 CC04 DD03 DD13 EE02                       EE03 EE24 EE37 FF24 FF26                       FF28 FF33 FF34

Claims (8)

[Claims] 1. An image pickup means for picking up an image of a lead group of an electronic component, a model shape information storage means for storing model shape information of a plurality of lead groups of the electronic component, and a plurality of lead groups obtained by the image pickup means. An image processing device having an image processing means for processing an image signal of an electronic component having the image processing means, wherein the image processing means outputs position information of a plurality of model feature points regarding a target lead group output from the model shape information storage means. The positional information of a plurality of feature point candidates of the lead group obtained from the image signal is compared, the combination of the feature point candidates closest to the relative positional relationship of the model feature points is selected, and the selected feature point candidates are selected. An image processing apparatus, wherein the position of the target lead group is calculated from the position information of the combination. 2. The image processing device according to claim 1, wherein the position information of the model feature point of the electronic component is calculated from the position coordinates of the brightness change point when the detection straight line is set in the predetermined direction of the target lead group. The position information of the feature point candidate is calculated from the position coordinates when the brightness change point of the target lead group is detected by applying the detection straight line to the image signal. 3. The image processing apparatus according to claim 2, wherein a combination of two points estimated to correspond to the two points having the largest distance among the plurality of model characteristic points is selected from the characteristic point candidates. The image processing apparatus is characterized in that the positions of the model feature points and the feature point candidates are compared and image processing is performed only for a combination of feature point candidates including the two points. 4. The image processing apparatus according to claim 3, wherein at the time of position comparison between the model feature point and the feature point candidate,
The combination of feature point candidates in which there is no brightness change point near the position estimated from the position information of the model feature points on the detection line of the image signal is judged as an incorrect combination, and the subsequent comparison calculation is not performed. An image processing device characterized by the above. 5. An image pickup means for picking up an image of a lead group of an electronic component, a model shape information storage means for storing model shape information of a plurality of lead groups of the electronic component, and a plurality of lead groups obtained by the image pickup means. Image processing means for processing an image signal of an electronic component having the image processing means, wherein the image processing means outputs position information of a plurality of model feature points regarding the target lead group output from the model shape information storage means and the image signal. The position information of the plurality of feature point candidates of the lead group obtained from is compared, the combination of the feature point candidates closest to the relative positional relationship of the model feature points is selected, and the combination of the selected feature point candidates is selected. An image processing method, wherein the position of the target lead group is calculated from position information. 6. The image processing method according to claim 5, wherein the position information of the model feature points of the electronic component is calculated from the position coordinates of the brightness change point when the detection straight line is set in the predetermined direction of the target lead group. The position information of the feature point candidate is calculated from the position coordinates when the brightness change point of the target lead group is detected by applying the detection straight line to the image signal. 7. The image processing method according to claim 6, wherein a combination of two points estimated to correspond to the two points having the largest distance among the plurality of model characteristic points is selected from the characteristic point candidates. An image processing method is characterized in that the positions of the model feature points and the feature point candidates are compared and image processing is performed only for a combination of feature point candidates including the two points. 8. The image processing method according to claim 7, wherein at the time of position comparison between the model feature point and the feature point candidate,
The combination of feature point candidates in which there is no brightness change point near the position estimated from the position information of the model feature points on the detection line of the image signal is judged as an incorrect combination, and the subsequent comparison calculation is not performed. An image processing method characterized by the above.