JPH03225212A - parts recognition device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a component recognition device, and in particular, by processing image No.
This invention relates to a device that extracts the shape of an item and recognizes the position of that part.

(Prior art) In order to check the mounting state of printed electronic components, a device as shown in FIG.
In figure 1 proposed in Publication No. 8206, (1)
(2) is a board, for example a printed circuit board, (2) is a part to be tested whose side part is 11, for example an electronic component, (3a), (3b)
is a light source that irradiates light from two directions so as to form shadows on different sides of the electronic component (2);
) is the light source (3a1. (3bl)
A photoelectric converter that captures an image of one component and converts it into an electrical signal, such as a Brevi camera, (5a), (5bl is the first and second converter that stores the image signal output by the photoelectric converter (4)).
The image memory (12) is an image processing device for image processing, which includes a subtraction device (11), an arithmetic device (6), and a 0 determination device (7).
). The light source (3a), (3bl and photoelectric converter (4) are connected to the light source (3a1. (3b) and photoelectric converter (
4) is arranged so as to be inclined with respect to the arrangement direction of the electronic component (2).

1 Dome's device is configured as described above, and detects the position and shape of parts by comparing the obtained image with a reference image. That is, first, the part to be inspected (2)
, turn on one light source (3a), and inspect the part to be tested (2).
irradiate. Next, an image of the light source surface is captured by a television camera (4), converted into an electric signal, and stored in the first image memory (
The zero-order light source (3a) to be stored in 5a) is turned off, the other light source (3b) is turned on, and an image of the light and heat radiation surface is captured by the television camera (4). This is converted into an electrical signal and stored in the second image memory (5b). The difference between the rising months 14 of the corresponding pixels of both image signals stored in these is calculated using a subtraction device (n111), and a difference image is obtained. From this difference image, the position of the part (2) is determined by the P1 calculating device (6), the result is manually input to the determining device (7), stored in advance, and compared with the rX value to determine flJ.

[Problem to be solved by the invention] In the conventional component recognition device as described above, the 1?
The shadows of parts in bright positions are also visible with the soldering bat in the background.
There is no bat with 1-10 on the fl, and it is 211 at the 118th position.
-The shadow of the item is also processed at the same time, and f? Since shadows must be extracted using the same threshold value from images with a brightness of f71 of 4, there is a problem that the range in which A hri can be set is narrow and the recognition rate is poor.

The present invention has been made to solve these problems, and an object thereof is to provide a component recognition device that can accurately and reliably recognize the shape and position of a component.

[Means for Solving the Problems] A component recognition device according to the present invention has a side part: 11 and 23 or more light sources capable of forming shadows on different sides of the product; It is equipped with a converter for converting the image of the part into an electric number 19, an image memory for storing the image signal outputted by the photoelectric converter, and an arithmetic unit for recognizing the side part of the part. A set of opposing sides of the product is recognized, a dj area surrounded by these sides is used as a window, and the inside of the window is treated as a processing 6n area and other sides are recognized.

[Function] In the present invention, from the image of the parts irradiated with the light source and stored in the image memory, firstly, from the image of the parts irradiated with the light source, a pair of facing parts that are likely to have a contrast with the background, i.e., 1'-E
By finding the side of the part where the TI pad is located, processing only the area determined by the two sides obtained from the results, and finding the other opposing side where it is difficult to create a contrast with the back. - Grade 16, shape or other geometry? It recognizes the characteristics of

〔Example] An embodiment of the present invention will be described below with reference to FIG.

FIG. 1 is a block diagram showing the configuration of a component recognition device according to an embodiment of the present invention. This device is a mounting condition inspection device that recognizes components mounted on printed circuit boards.
In the figure, (the eye is the pudding l-JA board, (2) is the part to be tested with a side part ■, which is mounted on a printed circuit board (!). (3) is the part to be tested (2) that is irradiated. It is a light source that
(:(a), (3b), (3e), (3
In d), the printed base & (1) i are installed at different angles, for example, the light source (3a) and the light source (3c) are installed diagonally so that they are symmetrical with the part to be inspected (2) in between, and the light source ( 3b)
are installed at 90 degrees horizontally from the light source (3a) and the light source (3c), respectively, and the angle and distance from the component (2) are ':9·Ft. Similarly light source (3d)
is installed at i'1 so as to sandwich the part to be inspected (2) and be symmetrical with the light source (3b), and the light source (3a) and the light source (3
They are installed at 90 degrees in the horizontal direction from c), and at the same angle and distance from component (2). That is, four light sources 13a), [3b), (
3c) and (3d) are arranged so as to form shadows on different sides of component (2), and when viewed from directly above component (2), they form an angle of 90°; Two sets are installed at opposing positions with component (2) in between. (
4) is light #(3al, (3bl, (3c),
A photoelectric converter that captures an image of the side of the component (2) by irradiation (3d) and converts it into an electrical signal, and is, for example, a television camera. (5) is the first image memory (5a) that stores the image signal output by the photoelectric converter (4), for example, the image signal when the light source (3a) is turned on yearly, and when the light source (3b) is irradiated. a second image memory (5b) that stores image signals of
, the third one stores the image signal when the light source (3C) is irradiated.
Image memory (5C), a fourth image memory (5d) that stores the image signal when the light source (3d) is irradiated, (b)
is an arithmetic device that calculates the maximum hfi image and the difference image to recognize the edge of part (2), and (7) is an arithmetic device that outputs and displays a judgment signal from the shape (>i) calculated by the arithmetic device (6). It is a judgment device.

A description will be given of JJI's component recognition device configured as described above. First, the printed circuit board +1) is placed at a position where it can be most imaged by the photoelectric converter (4), and the light source (31) is turned on to illuminate the printed circuit board [11]. This image is captured by the photoelectric converter (4) and stored in the first image memory (5a).
). The image at this time is shown in Figure 2 (a).
This light source (3a) is on the left when facing 119 items (2).
The shadow (8a) of the component is shown by diagonal lines in the figure. This part (2) has four sides (
2al, (2bl, (2c), (2
dl shall be its own.

Next, the light source (3d) is turned off, and an image of the irradiated surface is captured by the photoelectric converter (4) with the light source (3b) turned on from the direction F toward the component (2), and the second image memory (5b ). The image at this time is shown in Figure 2 (0-order light source shown in bl)
3b) and turn off the light source (-) from the right towards part (2).
3C) is turned on, an image of the Q-j plane is captured by the photoelectric converter (4), and is stored in the third image memory (5C). The image at this time is shown in Figure 2 (C) from the zero-order light source (3
c) is turned off by 11 degrees, and the light source (
3d) is turned on, the photoelectric converter (4) is the image of the irradiated surface.
) and store it in the fourth image memory (5d). The image at this time is shown in FIG. 2(d).

Next, a case will be described in which the side portion (2b) of the component (2) is recognized. For example, the first image memory (
5a) Compare the signal values corresponding to each pixel in the image stored in the third image memory (5C), and choose the one with the larger signal level representing one bright ship! Perform calculations and create a maximum of one direct image. This is shown in Fig. 3(a). In this maximum value image, a shadow (8a) of the part was created in the image where the part (2) was illuminated by the light source (3a): 11-min and the light source. By (3c), the part where the shadow (8C) of the part was created in the irradiated image of the part (2) is erased, and a part image without a shadow can be obtained in the processed image (13).0 Part (2)L At this time, the unevenness that can occur can also be removed.
The difference from the signal 111 stored in the second image memory (5b), which is the image of the shadow on the side to be recognized (2b), is determined and used as a difference image. This is shown in FIG. 3(b).

In other words, for the image in Figure 2, calculate the maximum value of the image signal ``a (x, yl) when irradiated with light source (3a) and the image signal rc (x, yJ) when irradiated with light source (3C). The maximum value and the light source (3b) are obtained using the device (6).
) Image signal fb (x, y) when illuminated by q・1
Find the difference between the two using the performance device (6). This is expressed as g(x, yl :maxtra(x, y)Jc(x,
yll-rb(x, yl.

Here, when we examine the difference image g (x, y) shown in Fig. 3 [b], we find that the second (maximum ft) shown in Fig. 3 [a] (when compared with the (8 value) of each pixel in the 4 images, Light source (3b)
In the image obtained by irradiating the part (2) according to the above, the part of the part where the light (8b) was formed is dark. Therefore, the signal f+a at the corresponding pixel is small,
The value of g(x, yl is positive 11α and remains as a shadow (8e).However, in the image of the part (2) illuminated by the light source (3a), the part where the shadow (8a) of the part is formed,
In the image of component (2) irradiated with light # (3C), the part where the shadow (8c) of the component is formed is the maximum (as it is erased when obtaining the direct image, there is no shadow in the processed image (13)). At this time, it is also possible to remove unevenness that appears on the 0 parts from which the part image can be obtained.
) (Figure 2(b))
The parts that are causing random specular reflection on the part are the image of the part (2) illuminated by the light source (3a) (Fig. 2 (a)) or the part (2) illuminated by the light source (3C).
Since similar specular reflection occurs in any of the images (Figure 2 (C)) irradiated with
It remains in Figure (a)). Therefore, the difference image (Fig. 3(b)
), the specular reflection part is erased. In this way, only the shadow portion (8e) of the side portion (2b) of the component can be extracted.

Next, the boundary line on the component side in this Akira (8e) is obtained, for example, by converting the image into two parts.For example, in FIG. 3(b),
Extract points that change from black to white when scanning vertically from the top to the bottom of the diagram.

That is, g (x, y) is extracted along the y axis at points that change from black to black in the upper half of the figure, and is linearized by applying Hough transform. Similarly, from the shadow (8d) formed on the lower side of the part shown in Figure 2 (dl), the lower side of the part (2
Recognize d). Now the opposite 1 on the %1 side of the part
The edges (2b) and (2d) of the set were recognized.

Next, a process of recognizing another pair of sides (2a), (2c) on the long sides of the part (2), the long side sides (2al, 2c), is performed. (2
el O) (T (F: Range is the short side (2
Since the processing range can be limited to the area between b) and (2d), the image within that range is filtered and the short side (
2b1. (Similarly to 2dl, obtain the maximum value image and the difference image, and further recognize the edges. At this time, the binarization threshold t+
The parameters of ri etc. are the short side (2b), (2dl
By changing the parameters used when calculating , the long sides (2a) and (2cl) can be recognized reliably and accurately.

After the test, the judgment device (7) calculates the calculation device 1: (The digit 16 or 1 inversion angle calculated in (6) and other data such as No. d are F.
The book is compared with the standard 11α, which is a preset standard, and judged as a good book, and the judgment result is displayed. Note that this determination device (7) is
What group is stored for? ¥ indicates the basic size of the parts, 8
What are the allowable deviations Ixt and allowable rotations t' of the 14 products? It is.

1. The operation described above will be explained by focusing on the image processing range. In FIG. 4(a), the processing area (13a) indicates the range where image processing is performed in the initial state, and this range is the part (
2) may be installed. Here, a pair of opposing short sides (2b1.+2d) is processed to extract a side part (14) as shown in FIG. 4(b). The area surrounded by this pair of sides (14) is defined as a window (+3b), and this window (13b) is the fourth
The processing range for recognizing the long sides (2a) and (2c) of the 0 component indicated by dotted lines in Figure fc) is within the window (13b), and within this processing range Tl3b), the long sides ( 2al, (2c) is recognized.

If the processing area i11 is limited as shown in FIG. 4(C), even if noise (15) is present, for example, it can be ignored as outside the processing range. That is, by reducing the influence of noise around the component body, the position, shape, or other geometrical features of the component can be reliably recognized. or,;? The threshold value can be set according to the background condition of the +L product (2), which widens the range in which the threshold f+Ii can be set, allowing stable and highly accurate recognition of the position, shape, or other geometric features of the part. be able to. Also, the side i1t in the direction that is easy to extract is recognized first, and the other sides are defined as the area (1
3b), the position i6, shape, or other geometrical features of the part can be reliably recognized.

In addition, in the embodiment described in -, the difference image is processed by the arithmetic unit (6) at 21
Convert to +rj to obtain a binarized image, perform 2 (8) conversion only to the edges of the image product side of the +iI image, and create part (2).
By straightening candidate points on the side (2d) of and finding the boundary between the part (2) and the background, the position iri of the part (2),
Although the configuration is configured to recognize shapes or other geometric features, the recognition is not limited to those that can be recognized without Hough transform.

Furthermore, the number of light sources is not limited to four, and the positions where they are installed do not have to be one at each 90° angle as in the above embodiment. , and may be arranged so that its sides can be easily recognized.

[Effects of the Invention] As described in -1- below, according to the present invention, it is possible to form shadows on different sides of a component having edges: S or more light sources, and shadows can be formed on different sides of the component by irradiation from the light sources. Side: Equipped with a photoelectric converter that captures an image of 1t and converts it into an electrical signal, an image memory that stores the image signal output by the photoelectric converter, and an arithmetic unit that recognizes the side of the part, the side 1 facing the part is equipped with a By recognizing the side of the set, using the area surrounded by this side as a window, and recognizing the other sides as the processing area within the 5f window, the parts can be positioned stably, reliably, and with high precision. This has the effect of providing a component recognition device that can recognize shapes and other geometric features.

[Brief explanation of drawings]

Fig. 1 is a block diagram showing the configuration of a parts recognition device according to an embodiment of the present invention, and Fig. 2 (al to (d)) relate to an embodiment of the present invention. An explanatory diagram showing the image, namely, Fig. 2 (at is the image when the part is irradiated with the light source (3a), Fig. 2 (bl is the image when the part is irradiated with the light source (3bJ), Fig. 2 (c) is the image when the part is illuminated with the light source (3c), and Figure 2 (d) is the image when the part is irradiated with the light source (3(1)
FIG. 1 is an explanatory diagram showing an image when a component is irradiated in step 1; Third
Figure (a) is an explanatory diagram showing a maximum of 116 images according to this embodiment, Figure 3 (bl is an explanatory diagram showing difference images, and Figure 4 (a)
) is an explanatory diagram showing the image processing range in the initial state according to this embodiment, and FIG.
Figure 4 ((
) is an explanatory diagram showing the image processing range limited from the result of FIG. 4(b), and FIG. 5 is a block diagram showing the configuration of a conventional component recognition device. +2) --One part, (2a), (2b),
(2c), (2d)...Side part, (3a1.
(3b1. (3c), (3dl... light source, 4
)---optical converter, (5a), (5b), (
5c1. (5dl...Image measurement, (6)...Arithmetic device, (7)...Judgment device, 18a), (8
b), 18c), (8d), [8e) (...Shadow of parts, (13al, (13b)...
- Processing area, (14)...Side portion of the linearized part. In addition, in the figures, the same reference numerals indicate the same or equivalent parts.

Claims (1)

[Claims] three or more light sources capable of forming shadows on different sides of a component having sides; a photoelectric converter that captures an image of the side of the component by irradiation with the light source and converts it into an electrical signal;
It is equipped with an image memory that stores the image signal outputted by the photoelectric converter, and an arithmetic unit that recognizes the sides of the component, and recognizes a pair of opposing sides of the component and calculates the area surrounded by the sides. A component recognition device configured to use a window as a processing area and to recognize other side parts using the inside of the window as a processing area.