JPH11289200A - Position recognition method and device - Google Patents

Abstract

(57) [Summary] [Problem] To automatically select an optimal pattern extraction method for an arbitrary combination of a substrate and a mark and recognize the position appropriately. An imaging system includes an imaging system for imaging an object to be inspected, and an image processing unit for storing a plurality of pattern extraction methods for processing a signal output from the imaging system and extracting a characteristic pattern. In the position recognition device, quantitative analysis means 2 and 3 for quantitatively analyzing the luminance level of the signal of the characteristic pattern and the luminance level of the signal of the background, and control for selecting the pattern extraction method of the image processing means 5 based on the result of the quantitative analysis. Part 4
And determining a luminance distribution state between the regions based on the result of the quantitative analysis of the luminance level of the image for each of the characteristic pattern shape region and the background region, and extracting an appropriate pattern according to the luminance distribution state of the image. I chose a method.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position recognition device suitably applied to a device for manufacturing an electronic component mounting board used in electronic equipment.

[0002]

2. Description of the Related Art A position recognition device in a conventional electronic component mounting board manufacturing apparatus will be described with reference to FIG.
In order to mount the component at a predetermined position on the board, the relative positional relationship between the device and the board is measured by recognizing the position of the mark which is a characteristic pattern provided on the board,
Attach the part to a predetermined position. Normally, the mark position is recognized by processing the grayscale information of the luminance by a pattern extraction method using the imaging system 1 and the image processing means 5 provided in the manufacturing apparatus, so that the mark shape provided on the substrate is processed. An extraction method is used.

[0003] The pattern extraction method for position recognition used in the image processing means 5 allows an operator operating a manufacturing apparatus to visually check the brightness and noise of an image obtained by capturing the characteristic pattern, the degree of irregularities in the pattern and the background, and the like. Through the input device 7, a method that is determined to be optimal and is determined to be optimal from among many pattern extraction methods registered in the control unit 4 in advance.

[0004] As another method, an operator who operates the manufacturing apparatus sets the material of the board on which components are mounted in the manufacturing apparatus of the electronic component mounting board and the material of the mark which is a characteristic pattern provided on the board. , Input to the control unit 4 via the input device 7, and selects a pattern extraction method corresponding to the substrate material and the mark material registered in advance in the control unit 4.

[0005]

When a mark provided on a substrate is imaged by an optical system and converted into luminance information, the luminance distribution state of the obtained image depends on the material of the mark to be imaged and the substrate. It varies depending on the material, the shape of the object, the magnification, the resolution, and the illumination state of the optical system used in the manufacturing apparatus. The reason why the luminance distribution state of the image changes depending on the mark material and the substrate is that the light reflectance is specific to the substance.

As described above, since the luminance distribution state of an image changes depending on many factors, in an electronic component mounting board manufacturing apparatus, when extracting a pattern for position recognition corresponding to each board, an optimal extraction is performed. In addition to the difficulty in selecting a method, this is largely due to the judgment of a skilled person having specialized knowledge of an image obtained by imaging the target substrate and the mark on the substrate and image processing.
For this reason, there has been a problem that it takes time to train workers who can select a pattern extraction method used for position recognition.

Further, a conventional method for classifying an image obtained by imaging a mark with an optical system based on a combination of the material of the substrate and the material information of the mark formed on the substrate and extracting a corresponding pattern for position recognition. A method of selecting the mark is also used, but if the forming method is different even if the mark material is the same, the imaging result of the formed mark is also different. For example, in the case where the quality changes before and after sintering as in the case of a ceramic substrate, the light reflectance before and after sintering differs. Also, as in the case where the three-dimensional shape greatly differs depending on the amount of solder such as a solder leveling mark, the way of reflecting light differs depending on the shape even if the material is the same, so that the luminance distribution state of the image obtained by imaging is It will be completely different. Therefore, there is a problem that an image cannot be classified only by a combination of materials.

[0008] Thus, when the operator erroneously selects the pattern extraction method for positioning, when the pattern extraction method selected by the operator does not correspond to an object, or when the material of the substrate or the mark material is used. When a pattern extraction method for position recognition corresponding to an image cannot be set due to inability to classify, if the production apparatus is operated and production is performed, a recognition failure occurs, and the production apparatus suspends operation and performs production. There is a problem that the component cannot be mounted at a predetermined position or is mounted at an incorrect position without mounting the component at a predetermined position, resulting in a defective product.

The present invention has been made in view of the above-mentioned conventional problems, and has as its object to provide a position recognizing apparatus capable of automatically selecting an optimum pattern extracting method for an arbitrary combination of a substrate and a mark and appropriately recognizing the position. The purpose is.

[0010]

According to the present invention, there is provided a position recognizing method for recognizing a position of a characteristic pattern provided for mounting a component at a predetermined position on a substrate. A second step of designating a feature pattern and a background area in the input image, a third step of calculating a standard deviation of the average luminance and the luminance distribution in each area, and a pattern based on the average luminance and the standard deviation of the luminance distribution. And a fourth step of selecting extraction. Even if it is not clear which pattern extraction method is appropriate for the inspection object, an appropriate pattern can be selected according to the luminance distribution state of the image. An extraction method can be selected.

Preferably, in the fourth step, a first evaluation value relating to contrast, which is a difference between average luminances of the characteristic pattern region and the background region, and a second evaluation value relating to a standard deviation of luminance distribution of both regions. An evaluation value is obtained, and a pattern extraction can be selected by plotting the evaluation value in a determination space having a vertical axis and a horizontal axis.

More specifically, the first evaluation value is large,
When the second evaluation value is small, pattern extraction is performed by the edge detection method. When both the first evaluation value and the second evaluation value are large, the edge detection window is arranged so as to surround the feature pattern in a circle. Pattern extraction is performed by the edge detection method, and when both the first evaluation and the second evaluation value are small, pattern extraction can be performed by the edge detection method that accumulates luminance levels and emphasizes boundaries.

[0013] A first part for reading component mounting data in which a plurality of pieces of information indicating the characteristics of the substrate and the pattern on the substrate, including the material of the substrate to be inspected and the material and forming method of the characteristic pattern on the substrate, is written. An image based on a combination of a plurality of pieces of information indicating the features of the substrate and the pattern on the substrate with reference to the storage means holding the combination of the process and a plurality of pieces of information indicating the features of the substrate and the pattern on the board and a pattern extraction method Even with the second step of selecting the pattern extraction method of the processing means, an appropriate pattern extraction method can be selected.

Further, the position recognition apparatus of the present invention has an image pickup system for picking up an image of an object to be inspected, and an image holding a plurality of pattern extraction methods for processing a signal output from the image pickup system and extracting a characteristic pattern. And a quantitative analysis means for quantitatively analyzing the luminance level of the signal of the characteristic pattern and the luminance level of the background signal other than the characteristic pattern, and a pattern of the image processing means based on the result of the quantitative analysis. A control unit for selecting an extraction method, quantitatively analyzing the brightness level of the image for each of the region of the characteristic pattern shape and the region of the background, and using a plurality of evaluation values based on the result of the quantitative analysis. By judging the luminance distribution state between the regions, even if it is unclear which pattern extraction method is appropriate for the inspection object, the luminance distribution state of the image can be determined. Flip it can be selected suitable pattern extraction method.

A storage means for storing a combination of a plurality of pieces of information indicating the characteristics of the substrate and the pattern on the substrate, including the material of the substrate to be inspected, the material of the pattern on the substrate and the forming method, and a pattern extraction method. And a control unit for selecting a pattern extraction method of the image processing means based on a combination of a plurality of pieces of information indicating characteristics of the substrate and the pattern on the substrate. By using a plurality of pieces of information, an image can be classified and a corresponding pattern extraction method can be selected.

[0016]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS One embodiment of the position recognition device of the present invention will be described below with reference to FIGS.

In FIG. 1 showing the configuration of the position recognition apparatus of this embodiment, a signal output by an imaging system 1 including an optical device for imaging an object to be inspected is quantitatively analyzed for quantitatively analyzing a characteristic pattern area. It is input to the means 2 and the quantitative analysis means 3 for quantitatively analyzing the background area other than the characteristic pattern. Further, the quantitative analysis means 2 and the quantitative analysis means 3 output a standard deviation representing the average luminance and the luminance variation of the characteristic pattern area and a standard deviation representing the average luminance and the luminance variation of the background area, respectively. The results output from the quantitative analysis means 2 and 3 are input to the control unit 4. The control unit 4 converts the average luminance and the standard deviation between the characteristic pattern and the background regions as described later, and further plots the evaluation value based on the average luminance and the standard deviation for each region on the characteristic space, thereby correspondingly. The information indicating the pattern extraction method to be performed is output to the image processing means 5. When performing a pattern extraction process for positioning, the image processing unit 5 sets a pattern extraction method based on information output from the control unit 4 and performs the process.

FIG. 2 shows an example of various pattern extracting methods.
FIG. 2A shows a case where the brightness level changes stepwise at the boundary between the mark area and the background area, as in the case where a solder mark on a glass epoxy substrate is imaged, and the magnitude of the brightness variation in each area. Is smaller than the magnitude of the change that rises stepwise at the boundary. As a pattern extraction method for positioning in the case of an image like this example, an edge detection (edge detection) method for detecting a stepwise change in luminance level at a boundary between a mark area and a background area is used. In this edge detection method, a variation in luminance in a mark area or a background area is detected in principle, but the magnitude at which the luminance level changes stepwise at each boundary is larger than the variation in luminance. Therefore, it is possible to remove the variation component of the luminance by using an appropriate threshold value.

FIG. 2B shows that the luminance changes stepwise at the boundary between the mark area and the background area as in the case where a gold-plated mark on an alumina ceramic substrate is imaged. This is a case where the magnitude is almost the same as the change in the luminance level in a stepwise manner at the boundary. In the case of this example, in the normal edge detection processing, the stepwise change of the luminance level at the boundary and the degree of the luminance difference due to the luminance variation in the mark area are almost the same, so that the luminance in the mark area It is difficult to set a threshold value for excluding the component of the variation, and the variation in the luminance may be erroneously detected as the boundary between the mark shape and the background. To solve this, a pattern extraction method is used in which an extraction window is arranged along the mark shape, an edge is detected in the extraction window, and an edge which is a boundary of the mark area is extracted.

FIG. 2C shows a case where there is almost no stepwise change in luminance at the boundary between the gold-plated mark area and the background area on the glass ceramic substrate, the luminance decreases at the outline of the mark shape, and the luminance varies in each area. is there. When the mark shape has only such a luminance change, the luminance level value of the image is accumulated in the horizontal or vertical direction to emphasize the drop in luminance at the outline of the mark shape, and the accumulated result is set to an appropriate threshold value. A pattern extraction method of extracting a contour by cutting with a.

Next, a method of selecting one of the above-described pattern extraction methods for position recognition will be described with reference to the flowchart of FIG. First, a captured image of a mark used for position recognition is input. Next, the area occupied by the mark to be extracted is specified on the image.

Next, a luminance distribution histogram is created in the area occupied by the mark, and a median M1 of the luminance distribution is calculated. Further, a standard deviation S1 representing the variation of the luminance distribution in the area occupied by the mark is obtained.

Next, similarly, a luminance distribution histogram of the background area is created, and a median M2 of the luminance distribution and a standard deviation S2 representing the variation of the luminance distribution in the background area are obtained.

Next, an evaluation value E1 representing the contrast between the mark area and the background area is obtained by using the equation (1).

E1 = | M1−M2 | (1) Further, an evaluation value E2 representing a variation in luminance between the mark area and the background area is obtained by using the equation (2).

E2 = (S1−S2) / (S1 + S2) (2) Next, the values are plotted in an evaluation space in which the horizontal axis represents the evaluation value E1 and the vertical axis represents E2, and the values are appropriately plotted in the evaluation space. The pattern extraction method is made to correspond by using the threshold curve, and the pattern extraction method corresponding to the plot position of the calculated evaluation value is output.

FIG. 4 shows an evaluation value E1 expressed by the equation (1) and an evaluation value E2 expressed by the equation (2).
A method for selecting the pattern extraction method of FIG. FIG. 4 (A)
FIG. 2A corresponds to FIG. 2A, in which the luminance level of the mark area and the background area changes stepwise at the boundary portion, and the variation in luminance in each area is smaller than the magnitude of the stepwise luminance change. When the evaluation value E1 is relatively large and the evaluation value E2 is small, this edge detection method can be applied.

FIG. 4 (B) corresponds to FIG. 2 (B), in which the luminance level of the mark area and the background area changes stepwise at the boundary, but the magnitude of the luminance variation in the mark area is stepwise. The circular window arrangement edge detection method can be applied when the evaluation value E1 is relatively large and the evaluation value E2 is large accordingly.

FIG. 4C corresponds to FIG. 2C, in which there is no stepwise luminance change between the mark area and the background area, the luminance decreases at the contour of the mark shape, and the luminance varies in each area. When both the evaluation values E1 and E2 are relatively small, an edge detection method for accumulating the luminance levels and enhancing the boundaries can be applied.

As described above, in each case, the evaluation value E1
By plotting the evaluation value E2 and the evaluation value E2 on the feature space in FIG. 4, it is possible to select a corresponding pattern extraction method.

Next, another embodiment will be described with reference to FIGS. In the present embodiment, a substrate / feature pattern information storage unit 6 is provided as shown by a virtual line in FIG. 1, and the correspondence between a combination of a plurality of pieces of information indicating the features of the substrate and the pattern on the substrate and the pattern extraction method is determined. Is held. The data in the substrate / characteristic pattern information storage means 6 is read out by the control unit 4. On the other hand, the data of the numerical information registered in the electronic component mounting board manufacturing apparatus includes a code as shown in FIG. 11 is a code representing the material of the substrate, 12 is a code representing the material of the characteristic pattern provided on the substrate, and 13 is a code representing the method of forming the characteristic pattern. By providing the code 13 representing the method of forming the characteristic pattern in this manner, the material is changed in quality before and after sintering, such as a ceramic substrate, so that the light reflectance as a material property value changes. Can also be classified.

In the present embodiment, the control unit 4 combines a plurality of pieces of information indicating the characteristics of the substrate and the pattern on the substrate held in the substrate / characteristic pattern information storage means 6 based on the respective codes of the target substrate. The pattern extraction method is read out from the correspondence between the pattern extraction method and the pattern extraction method and output. In this way, the image is classified by using a plurality of pieces of information on the features of the substrate and the pattern formed on the substrate,
A corresponding pattern extraction method can be selected.

[0033]

According to the position recognition method and apparatus of the present invention, as described above, the luminance level of the image is quantitatively analyzed for each of the characteristic pattern shape region and the background region, and a plurality of the luminance levels are determined based on the result of the quantitative analysis. By judging the luminance distribution state between the regions using the evaluation value of, even if it is not clear which pattern extraction method is appropriate for the inspection object, it is possible to determine the luminance distribution state of the image according to the luminance distribution state of the image. An appropriate pattern extraction method can be selected. This eliminates the need for an expert having image and image processing expertise to select a pattern extraction method, and eliminates the need for expertise in setting the pattern extraction method. And the time required to select a pattern extraction method for position recognition is reduced, and the number of defects generated in the process caused by the operator erroneously selecting the pattern extraction method is reduced. Can be

Further, storage means holding a combination of a plurality of pieces of information indicating the characteristics of the substrate and the pattern on the substrate, including the material of the substrate to be inspected, the material of the pattern on the substrate and the forming method, and the pattern extraction method. And a control unit that selects a pattern extraction method of the image processing unit based on a combination of a plurality of pieces of information indicating a pattern feature on the board and the board,
By using a plurality of pieces of information on the characteristics of the substrate and the pattern formed on the substrate, it is possible to classify the image and select a corresponding pattern extraction method.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an embodiment of a position recognition device of the present invention.

FIG. 2 is an explanatory diagram of each example of an image data and a pattern extraction method in the embodiment.

FIG. 3 is a calculation flowchart of an evaluation value used for selecting a pattern extraction method in the embodiment.

FIG. 4 is an explanatory diagram of each example of image data and determination processing in the embodiment.

FIG. 5 is an explanatory diagram of a data structure in another embodiment of the position recognition device of the present invention.

FIG. 6 is a block diagram illustrating a configuration of a conventional position recognition device.

[Explanation of symbols]

 REFERENCE SIGNS LIST 1 imaging system 2 quantitative analysis means 3 quantitative analysis means 4 control unit 5 image processing means 6 substrate / feature pattern information storage means

 ──────────────────────────────────────────────────続 き Continued on the front page (72) Inventor Yoichiro Ueda 1006 Kadoma Kadoma, Osaka Prefecture Matsushita Electric Industrial Co., Ltd.

Claims (7)

[Claims] In a position recognition method for recognizing a position of a characteristic pattern provided for mounting a component at a predetermined position on a substrate, a first step of inputting an image of the characteristic pattern, A second step of designating the background area, a third step of calculating the average luminance and the standard deviation of the luminance distribution in each area, and a fourth step of selecting pattern extraction based on the average luminance and the standard deviation of the luminance distribution. A position recognition method comprising: 2. In a fourth step, a first evaluation value relating to contrast, which is a difference between average luminances of both the characteristic pattern region and the background region, and a second evaluation value relating to a difference between standard deviations of luminance distribution of both regions. 2. The position recognition method according to claim 1, wherein a pattern extraction is selected by plotting in a decision space having both evaluation values as a vertical axis and a horizontal axis. 3. When the first evaluation value is large and the second evaluation value is small, pattern extraction is performed by the edge detection method.
When both the evaluation value and the second evaluation value are large, pattern extraction is performed by an edge detection method in which an edge detection window is arranged so as to surround the feature pattern in a circle, and both the first evaluation value and the second evaluation value are small. 3. The position recognition method according to claim 2, wherein pattern extraction is performed by an edge detection method that accumulates luminance levels and emphasizes boundaries. 4. A position recognition method for recognizing a position of a characteristic pattern provided for mounting a component at a predetermined position on a substrate, wherein the material of the substrate to be inspected, the material of the characteristic pattern on the substrate, A combination of a first step of reading component mounting data in which a plurality of pieces of information indicating features of a substrate and a pattern on a board including a forming method are written, and a plurality of pieces of information indicating the features of the board and a pattern on the board and a pattern extraction method. A second step of selecting a pattern extraction method of the image processing means based on a combination of a plurality of pieces of information indicating characteristics of the pattern on the board with reference to the held storage means. Method. 5. A position recognition system comprising: an imaging system for imaging an object to be inspected; and image processing means for holding a plurality of pattern extraction methods for processing a signal output from the imaging system and extracting a characteristic pattern. A quantitative analysis unit for quantitatively analyzing the luminance level of the signal of the characteristic pattern and a luminance level of a background signal other than the characteristic pattern; and a control unit for selecting a pattern extraction method of the image processing unit based on the result of the quantitative analysis. A position recognition device comprising: 6. A position recognition apparatus comprising: an image pickup system for picking up an image of an object to be inspected; and image processing means for holding a plurality of pattern extraction methods for processing a signal output from the image pickup system and extracting a characteristic pattern. In the apparatus, storage means holding a combination of a plurality of information and a pattern extraction method indicating the characteristics of the pattern on the substrate and the substrate including the material of the substrate to be inspected and the material and the formation method of the pattern on the substrate, A position recognition apparatus comprising: a substrate; and a control unit that selects a pattern extraction method of an image processing unit based on a combination of a plurality of pieces of information indicating characteristics of a pattern on the substrate. 7. An electronic apparatus manufacturing apparatus comprising the position recognition device according to claim 5.