JPH0458374A - Picture processor, inspection device for object and pattern recognition device - Google Patents

Abstract

PURPOSE:To calculate the position coordinate of objects with high accuracy by providing a data calculation means outputting more than one kind of picture calculation data while performing such as differential and synthetical processing or the like of more then two kinds of picture data. CONSTITUTION:A storage means 11, a data calculation means 12, a signal processing means 13, and a control means 14 are provided. The 'n' kinds of picture data Din1 to Dinn are respectively stored in the storage means 11 composed of storage means M1 to Mn, and more than two kinds of picture data Dini, Dinj, selected from the stored picture data Din1 to Dinn by the control means 14, are outputted to the data calculation means 12 as data to be calculated D1 to Dn. The functional calculation processing such as the differential and synthetic processing or the like of the data to be calculated D1 to Dn is performed by the data calculation means 12 to be outputted as more than one kind of picture calculation data d1 to dn. Thus, the position coordinate of the objects can be calculated with high accuracy from waveform falling position when performing the picture processing specifying the position, etc., of the object.

Description

[Detailed Description of the Invention] [Table of Contents] Overview Industrial Application Fields Prior Art (Fig. 13) Problems to be Solved by the Invention (Fig. 14) Means for Solving the Problems (Figs. 1 to 3) ) Functional example (i) Description of the first example (Figures 4 to 10) (j)
Description of the second embodiment (Fig. 11, Fig. 12) Effects of the invention [Summary] An image processing device, an object inspection device, and a pattern recognition device, in particular, acquire an image of an object to be inspected such as a semiconductor component, Regarding the device and the signal processing device that performs the inspection and recognition processing, the slice level is directly set to the gradation information and height information of the object to be inspected, and the slice level is not subjected to inspection processing, etc., but the function processing is performed on it. The purpose of the processing device is to increase the reliability of individual information and perform inspections with high accuracy, and the processing device has a storage means for storing nMM image data and outputting at least two or more types of operand data. , a data calculation means for performing functional calculation processing on the two or more types of operand data and outputting one or more types of image calculation data; and a signal processing means for performing signal processing on the image calculation data and outputting image processing data. , comprising a control means for controlling the input/output of the storage means 1 data calculation means and the signal processing means, and the inspection device is an apparatus equipped with the processing device,
The image processing device includes at least an object image acquisition unit that acquires an image of the object to be inspected and outputs the n types of image data, and first image processing data and determination criteria output from the signal processing unit. and a first determination means for inputting first reference data and outputting first determination result data, the recognition device being a pattern recognition device equipped with the processing device. The image processing device includes at least pattern image acquisition means for acquiring an image of the object to be recognized and outputting the nmi image data, second image processing data output from the signal processing means, and and a second determination means for inputting second reference data serving as a pattern determination standard and outputting second determination result data.

[Industrial application field]

The present invention relates to an image processing device, an object inspection device, and a pattern recognition device, and more specifically to a device and a signal processing device for irradiating an inspection target such as a semiconductor component with inspection light to perform inspection and recognition processing. It is related to.

2. Description of the Related Art In recent years, in the field of image processing technology for the purpose of factory automation, etc., there has been a demand for a device that non-contactly measures the three-dimensional shape of a component mounted on a printed circuit board, for example.

This is because by automating the parts inspection process and the recognition process, which were conventionally performed visually, it is possible to improve factory productivity and reduce costs.

By the way, according to conventional devices, the inspection processing is performed by performing image processing using three-dimensional information (object height) in addition to two-dimensional gradation information (object brightness) of the object to be inspected. There are some examples where this is done.

According to this, the S/N (signal-to-noise) ratio of two-dimensional gradation information and three-dimensional information decreases depending on the color and gloss of the object to be inspected, the illumination conditions of the image acquisition system, etc., resulting in accurate positioning. Detection may not be possible. For example, when an object is scanned and imaged and the falling point of the obtained signal is used to identify the position of the object's shape change, the position cannot be detected accurately, so subsequent inspection and recognition processing is difficult. There is a problem in that it cannot be performed with high precision.

Therefore, without directly signal-processing shading information and height information and performing inspection processing, etc., we perform functional calculation processing to increase the reliability of each piece of information and enable highly accurate inspection, etc. The device and its applied devices are desired.

[Conventional technology]

13th. FIG. 14 is an explanatory diagram related to the conventional example, and the first
FIG. 3 shows a configuration diagram of a conventional object inspection device.

In the figure, for example, an apparatus for non-contactly inspecting the quality of semiconductor components of a printed circuit board, which is an object to be inspected 9, includes an image acquisition means 11, an image input circuit 2.degree. height image memory 3. Brightness image memory 42 image processing circuit 5. It consists of a window generation circuit 61, a position determination circuit 7, and a CPU 8.

The function is, for example, when detecting the mounting position of the object to be inspected 9, the object to be inspected 9 is first irradiated with the inspection light L1 scanned by the optical scanning device IA of the image acquisition means 1.

At this time, the object to be inspected 9 is stage-moved in a direction substantially perpendicular to the scanning direction of the inspection light L1.

In addition, the reflected light L2 from the object to be inspected 9 is reflected by the imaging lens IB.
PSD (Po5ition 5en
sitive detector, position detection light receiving element) is input to the IC. The detection signal S from the PSDIC includes height information and brightness information of the object 9 to be inspected. These two types of information are converted into digital signals by the image input circuit 2, and the height image data DMI is stored in the height image memory 3, and the brightness image data DM2 is stored in the brightness image memory 4.
are stored respectively. In addition, both data DMI, DM
2 is a CPU 8 and a window generation circuit 6 on a memory area.
The inspection area is specified via the .

Furthermore, the data selection circuit 5A and CP of the image processing circuit 5
In the zero position search circuit 5B, in which one piece of data is selectively outputted to the position search circuit 5B by U8, for example, the slice level SL is set in the height image data DMI, and the mounting position of the object to be inspected 9 is detected, Its position detection data DP
O is output to the position determination circuit 7. In the determination circuit 7,
The design reference data DR and the position detection data are compared and verified, and the resulting data Do is output to the higher-level system.

Thereby, it is possible to inspect the quality of the mounting state of semiconductor components and the like without contact.

[Problem to be solved by the invention]

By the way, according to the conventional example, when detecting the position of the object 9 to be inspected, the slice level SL is uniquely set in either the height image data DMI or the brightness image data DM2. There is.

For this reason, as shown in the problem in Figure 14, for example, when image processing is performed to compare the actual separation distance from the tip position of the object with its separation reference value, an error may occur in the detected position amount. May intervene.

That is, in the cross-sectional view of the bond between the LSI wiring lead 9A and the printed circuit board 9B in FIG. Position coordinates (x, y) and height image data DMI or brightness image data D? Image processing is performed based on either one of the 12 data.

Here, it is assumed that the actual stage coordinate system is x, y, z, the actual tip position (2) of the lead 9A is Xl, and the actual protruding position of the solder joint 9c is X2.

In this image processing, a slice level SL is set to the waveform h (x) which represents the height image data DMI as an analog signal in FIG.
The stage coordinate position X1 of the falling position is processed and recognized as the tip position of the object. Here, assuming that the stage coordinate system of the image processing system is X, 3', and Z, the tip position 12=xl of the lead 9A is recognized.

In addition, the process for recognizing the protruding length l of the solder joint 9C from the tip position of the solder joint 9C from the wiring lead 9A is as follows.
The stage coordinate position x2 at the time L2 of the next waveform peak value is calculated from the waveform peak value recognized as the tip position ■ of the waveform i (x) representing the brightness image data DM2 in the same figure (c) as an analog signal. This is done by Similarly, at this time, the protrusion recognition position x2 of the solder joint 9C is
Suppose that we obtain

However, an error α may occur between the actual protrusion length fl=X2-XI and the perceived protrusion length 12=x2-xi. This is due to the color and gloss of the object to be inspected 8, the illumination conditions of the image acquisition system, etc., resulting in a decrease in the S/N (signal to noise) ratio of -dimensional gradation information and three-dimensional height information, and a significant difference in the slice level SL. This is thought to be due to a decrease in the reliability of individual data due to the setting process.

As a result, position detection cannot be performed with high precision, resulting in a problem that inspection accuracy and recognition accuracy are reduced.

The present invention was created in view of such conventional problems, and it is possible to directly set a slice level to the shading information and height information of the object to be inspected and process it by functional calculation without performing inspection processing. The present invention aims to provide an image processing device, an object inspection device, and a pattern recognition device that increase the reliability of individual information and make it possible to perform inspections, etc. with high accuracy.

[Means to solve the problem]

1 to 3 respectively show principle diagrams of an image processing device, an object inspection device, and a pattern recognition device according to the present invention.

The processing device processes image data Djn1 to Di of type n11.
nn and outputs at least two or more types of operand data DI=Dini, Dn=Dinj; The above image calculation data d1 to d
data calculation means 12 that outputs the image calculation data d1 to dn, signal processing means 13 that performs signal processing on the image calculation data d1 to dn and outputs the image processing data Dout, and the storage means 11. The processing device is characterized by comprising a control means 14 for controlling input/output of the data calculation means 12 and the signal processing means 13, wherein the storage means 11 stores the first to n-th image data Din1 to Memorize Dinn individually and store the 1st to nth
1st to nth outputs individually the operand data D1 to Dn of
It consists of storage means M1 to Mn, and the data calculation means 1
2 is at least the first to nth image data Din
The inspection device is characterized in that it performs differential processing or synthesis processing of two or more types of operand data D1 to Dn selected from 1 to Dinn, or the differential processing and synthesis processing. An object inspection device comprising:
The image processing device acquires at least an image of the object to be inspected 19 and stores the n types of image data Din1 to Din.
object image acquisition means 15 that outputs n, and first image processing data Dout outputted from the signal processing means 13;
1 and a first determination means 16 that inputs first reference data DRI serving as a determination criterion and outputs first determination result data DAI, the inspection apparatus comprising: The data calculation means 12 at least
performing n-th differential processing on image data Dini selected from the first to n-th image data Din1 to Dinn;
The recognition device is characterized in that a multiplication process is performed on the data subjected to the n-th differential processing, and the recognition device is a pattern recognition device equipped with the processing device, and the recognition device includes at least an image of the object to be recognized 20 in the image processing device. a pattern image acquisition means 17 that acquires and outputs the n types of image data (Din1 to Dinn); second image processing data Dout2 output from the signal processing means 13; The present invention is characterized in that it is provided with a second determining means 18 that inputs the reference data DR2 and outputs the second determination result data DA2, thereby achieving the above object.

[For production]

According to the processing device of the present invention, storage means 11. Data calculation means 12. Signal processing means 13 and control means 14 are provided.

For example, first to nth n types of image data Din1 to D
inn is individually stored in the storage means 11 consisting of the first to nth storage means M1 to Mn, and two or more types of image data selected by the control means 14 from among the stored n types of image data Dia1 to Dinn are stored. Image data Dini, Dinj
are outputted to the data calculation means 12 as operand data D1 to Dn. The two or more types of operand data D1 to Dn are subjected to differential processing or synthesis processing by the data calculation means 12,
Alternatively, functional calculation processing such as differential processing and composition processing is performed, and the resultant data is output as one or more types of image calculation data d1 to dn.

Therefore, even if image processing is performed to detect the separation distance from a certain position of the image acquisition target to an arbitrary position as in the conventional example, two or more types of operand data D1 to Dn
For example, the operand data D1 to Dn of two or more types of coughs are subjected to functional calculation processing such as differentiation processing and composition processing by the data calculation means 12 without uniquely setting the slice level S17 to any one of the data. By doing this, when performing image processing to specify the position of an object, etc., it becomes possible to accurately calculate the position coordinates (X, Y) of the object from the falling position of the waveform.

This improves the reliability of individual data, and by using the highly reliable data for subsequent signal processing,
It becomes possible to improve the inspection accuracy and recognition accuracy.

Further, according to the inspection device of the present invention, the processing device is provided with object image acquisition means 15 and first determination means 16.

For example, if the image of the object to be inspected 19 is
, and are individually stored as the first to n-th n types of image data Din1 to Dinn in the storage means 11 consisting of the first to n-th storage means M1 to Mn. In addition, n types of stored image data Din1 to Dinn
Two or more types of operand data Dini selected by the control means 14 from the following.

Dinj is output to the data calculation means 12. moreover,
Two or more types of operand data D1 to Dn are first subjected to -order differential processing by the data calculation means 12, and then,
A functional operation process such as a multiplication process is performed on the −th order differential processed data. It is one or more types of image calculation data d1 to d
It is output to the signal processing means 13 as n. In the signal processing means 13, one or more types of image calculation data d1 to dn are subjected to signal processing and outputted to the first determination means 16 as first image processing data Dout. The first image processing data Dout is, for example, compared with first reference data DR1 serving as a determination standard by the first determining means 16 and output as first determination result data DAI.

For this reason, even if the position of the object to be inspected 19 is detected by image processing that compares the actual separation distance from the tip position of the object to be inspected with the separation reference value as in the conventional example, there are two types of mounting methods. Without uniquely setting the slice level SL to any one of the above operand data D1 to Dn, for example, the two or more types of operand data D
1 to Dn are subjected to negative order differentiation processing by the data calculation means 12, and then functional calculation processing such as multiplication processing is performed on the data subjected to the negative order differentiation processing, so that the inspected object 19 is obtained from the falling position of the waveform. The position coordinates (X, Y) of can be calculated with high precision. As a result, the first image processing data Dout in which the intervention of errors is suppressed as much as possible
By performing a comparison process between the first reference data DRI set in advance and the first reference data DRI set in advance, it becomes possible to output the first determination result data DAI of the high confidence orbit.

This makes it possible to accurately inspect the mounting position of the object 19 to be inspected, such as a semiconductor component.

Furthermore, according to the recognition device of the present invention, the processing device is provided with pattern image acquisition means F and second determination means 18.

For example, an image of the object to be recognized 20 is acquired by the pattern image acquisition means 17, and it is individually stored as the first to nth n types of image data Din1 to Dinn from the first to nth storage means M1 to Mn. The information is stored in the storage means 11 consisting of: Also, the stored n types of image data Dinl to Dj
Two or more types of operand data Dini and Dinj selected by the control means 14 from among the data calculation means 12
is output to.

Furthermore, two or more types of operand data D1 to Dn are combined by the data calculation means 12. The signal processing means 13 converts it into one or more types of image calculation data d1 to dn.
is output to. In the signal processing means 13, one or more types of image calculation data d1 to dn are subjected to signal processing,
As the image processing data Dout, the second determination means 18
is output to. The second image processing data Dout is, for example, compared with second reference data DR2 serving as a determination standard by the second determining means 18, and outputted as second determination result data D as 2.

Therefore, even if pattern recognition is performed by image processing that compares brightness image data and sadness image data of the object to be recognized 20, two or more types of operand data D
Slice level SL for any one of data 1 to Dn
For example, the two or more types of operand data D1 to Dn are subjected to correlation processing by the data calculation means 12 without performing unambiguous number setting processing, and then a function such as synthesis processing of the correlated data is performed. By performing the arithmetic processing, it is possible to accurately recognize the positional shift of the object to be recognized 20 based on the sharpness of the matching degree. As a result, by performing a comparison process between the second image processing data Dout in which the intervention of errors is suppressed as much as possible and the second reference data DR2 set in advance, the second judgment result of high customer confidence is obtained. Data D
It becomes possible to output A2.

This makes it possible to accurately recognize the mounting condition of the object 20 to be recognized, such as a semiconductor component.

〔Example〕

Next, each embodiment of the present invention will be described with reference to the drawings.

4 to 12 show an image processing device according to each embodiment of the present invention,
FIG. 2 is an explanatory diagram of an object inspection device and a pattern recognition device.

(i) Description of the first embodiment FIG. 4 shows a configuration diagram of an image processing apparatus according to the first embodiment of the present invention.

In the figure, for example, an image processing device that processes two types of image data Dinl and Djn2 including the position coordinates (x, y) of an object to identify the position of the object is the first one. Second image memory 21A, 21B, data calculation circuit 22. Signal processing circuit 23. It consists of a central processing unit (hereinafter referred to as CPU) 242, an image input circuit 25, and a window generation circuit 26.

That is, 21A and 21B are image memories serving as an embodiment of the storage means 11, and two types of image data Dinl,
Din2 is individually stored, and two types of operand data DI and D2 are individually outputted by a control signal from the window generation circuit 26.

Reference numeral 22 denotes a data calculation circuit which is an embodiment of the data calculation means 12, including a differentiation circuit 22A, a -order differential value data memory (1) 22B, and a -order differential value data memory (U) 22C.
, a second-order differential value data memory 22D, and a multiplication circuit 22E. The differentiating circuit 22A individually differentiates two types of operand data D1 and D2, which is an example of functional operation processing, and outputs the results, and a -th order differential value data memory (1) 22B.

The -th order differential value data memory (I1) 22C is for individually storing the operand data DI, D2 (hereinafter referred to as -th order differential data di, d2) that has been subjected to the -th order differential processing.

Further, the differentiation circuit 22A further differentiates the differentiated first-order differential data d1 and outputs second-order differential data d3, and the second-order differential value data memory 22D stores it. . The multiplication circuit 22E is, for example,
The -th differential data d1 and the second differential data d3 are multiplied and outputted as image calculation data d4.

Reference numeral 23 denotes a signal processing circuit which is an embodiment of the signal processing means 13, and is used to process one image calculation data d4 subjected to function calculation processing.
This is to signal-process the data and output image processing data Dout.

The signal processing at this time is, for example, when the image calculation data d4 is an analog signal indicating the shape position where the object partitions the space, the falling position and the rising position of the waveform are detected, and the position of the object is determined from the detected position. This is to perform arithmetic processing on coordinates (X, Y).

24 is a CPU which is an embodiment of the control means 14, and the first. Second image memory 2
1A, 21B, the data calculation circuit 22, each differential value data memory 22B to 22D, and the input/output of the signal processing circuit 23.

25 is an image input circuit, which receives the first . Second analog image signal AinI A1n
2 is subjected to analog/digital processing, and the first and second image data Dinl and Din2 are individually output.

26 is a window generation circuit, which generates the first . Second image memory 21A 21B
It supplies an address to the differential circuit 22A and defines a data processing area and the like.

Next, the operation of the image processing apparatus according to the first embodiment of the present invention will be explained.

FIGS. 5(a) to 5(d) show operation explanatory diagrams of the image processing apparatus according to the first embodiment of the present invention.

In the same figure (a), for example, the position coordinates of the object 19A (
Two types of analog image signals A1n1. A
1n2 to perform image processing to identify the position of the object 19A, first, an image of the object 19A is acquired. Here, it is assumed that the object 19A is being moved and scanned in the X direction. Also, 1st. Second analog image signal A1n
1. A1n2. For example, assume that a brightness image signal i (x) and a height image signal h (X) are output from the sensor 27A.

Further, FIG. 2B shows the signal waveforms of the brightness image signal i (x) and the height image signal h(χ) before being subjected to analog/digital processing. In the figure, the vertical axis represents signal waveforms i (x) and h (x), and the horizontal axis represents the position X of the object 19A. For convenience of explanation, both signal waveforms are shown as analog waveforms even after analog/digital processing.

Here, Xl is the falling position of the object 19A, and x2 is its base position.

Therefore, the brightness image signal i (x) and the height image signal h (x) that have been subjected to analog/digital processing are individually divided into the first and second image data Dinl, Din2. The images are stored in the second image memories 21A and 21B. Next, the two types of stored image data Dinl and Din
2 is selected by the CPU 24 as the operand data DI, D2, and is output to the data calculation circuit 22.

The processing up to now is similar to the conventional example, but the present invention is different in the following points.

That is, the same figure (c) shows a signal waveform based on the -order differential data di, d2 and the second-order differential data d3 of the brightness image signal i (x) and the height image signal h (x) related to the image of the object 19A. are shown respectively.

In the same figure (c), the vertical axis is image calculation data and shows each differential data d1 to d3. The horizontal axis is object 19A
position X is shown. Also, dl is the brightness image signal i
(x) -th differential data, data calculation circuit 2
This is the result of differential processing using 2. Here, for convenience of explanation, the -th differential data d1 is the -th differential of both the signal i (x) and the signal h (x) as (i(x))' and (h(X)
)' and similarly, the second-order differential of the height image signal h (x) is represented as [h (x) )'.

These -order differential data di, d2 are individually stored in the respective differential value memories 22B, 22C, and the differentially processed first-order differential data d1 is further differentiated to become second-order differential data d3, which is a second-order differential data d3. It is stored in the differential value data memory 22D.

Figure (d) shows the −th order differential data d1 of the brightness image signal i (x) and the height image signal h (
It shows a composite waveform obtained by multiplying the second-order differential data d3 of x).

In the same figure (d), d4 is image calculation data, −th order differential data d1 of signal i (x) and signal h (x
) multiplication value dlXd3 of second-order differential data d3 = [i
(x) )' xi2 is the base position of the composite waveform, which is the base position x2 of the object 19A subjected to functional calculation processing.These are calculated by the multiplication circuit 22H, and the multiplied image calculation data d4 is sent to the signal processing circuit. 23.

In this way, according to the processing apparatus according to the embodiment of the present invention, the first. Second image memories 21A, 21B.

Data performance! Circuit 22. Signal processing circuit 23. CPU24,
An image input circuit 25, a window generation circuit 26, etc. are provided.

Therefore, even if image processing is performed to detect the separation distance from a certain position of the image acquisition target to an arbitrary position as in the conventional example, two or more types of operand data DI, D2
For example, the analog/digital processed brightness image signal 1(x), height image signal h
(x) is selected by the CPU 24 as the operand data DI, D2, and is output to the data calculation circuit 22.

In the data calculation circuit 22, -th differential data d1=(i(
Functional calculation processing such as multiplication processing of x))' and second-order differential data d3=[h(x))2 is performed. With this, the position coordinates 1 for specifying the position of the object, etc. from the falling position of the waveform subjected to image processing, for example, the X direction position x 11.
It becomes possible to calculate x 12 with high accuracy.

This improves the reliability of individual data, and by using the highly reliable data for subsequent signal processing,
It becomes possible to improve the inspection accuracy and recognition accuracy.

FIG. 6 shows the configuration of an object inspection device according to a first embodiment of the present invention.

In the figure, the device for non-contact testing of the quality of the semiconductor component 29A of the printed circuit board 29, which is the object to be tested 19, is as follows:
At least a three-dimensional shape acquisition device 2 is provided in the image processing device.
7. A first determination circuit 28 and a stage drive device 30 are provided.

That is, 27 is a three-dimensional shape acquisition device which is an example of the object image acquisition means 15, and acquires an image of the printed circuit board 29, for example, the first . Second analog image signal A
1n1. Brightness image signal i (x) that becomes A1n2,
It outputs a height image signal h(χ).

Reference numeral 28 denotes a first determination circuit which is an embodiment of the first determination means 16, and outputs from the rising edge detection circuit 23A and the falling edge detection circuit 23B which are embodiments of the signal processing circuit 23 described above. First and second image processing data Dout 1. Dout 2 and first reference data DRI serving as a determination standard are input, and first determination result data DAI is output.

In addition, 30 is a stage drive device, and the printed circuit board 2
9 is placed on a stage that moves and scans. Stage drive device 30 is controlled by a control signal from CPU 24.

Next, the operation of the inspection device will be explained.

7 to 10 are explanatory diagrams of the operation of the object inspection apparatus according to the first embodiment of the present invention, and FIG. 7 is an explanatory diagram of the object to be inspected according to the first embodiment of the present invention. ing.

29 is a printed circuit board to be inspected 19, and 29A is a semiconductor component mounted on the printed circuit board 29. X and Y are the stage coordinate system, and serve as the reference coordinates of the inspection apparatus. Here, it is assumed that there is a request to inspect the protrusion of the solder joint 29B from the tip 1 of the lead wire 29C of the semiconductor component 29A.

FIG. 8 is an explanatory diagram of image data according to the first embodiment of the present invention, and shows the relationship between a cross-sectional view of the semiconductor component 29A, height image data, and brightness image data.

In the figure, ■ is the tip position and indicates the boundary position P(xi) between the tip of the lead wire 29C of the semiconductor component 29A and the solder joint 29B. In addition, position P (x2)
is the position where the solder protrudes from the tip position (2), and the distance l thereof is P (x2) - P (x1).

When inspecting the printed circuit board 29 that includes such a positional relationship, first, an image of the semiconductor component 29A is acquired by the three-dimensional shape acquisition device 27. Here, the semiconductor component 29
It is assumed that A is being moved and scanned in the X direction. At this time, 1. Second analog image signal A1n1. Brightness image signal 1 (x) and height image signal h (x
) is output from the three-dimensional shape acquisition device 27. These two types of analog image signals A1n1. A1n2 is subjected to analog/digital processing by the image input circuit 25, and thereafter subjected to image processing to specify the position of the semiconductor component 29A, etc.

In other words, the brightness image signal 1(x) and the height image signal h are analog/digitally processed by the image input circuit 25.
(x) is the first. Second image data Dinl Din
2, it is individually the first. Second image memory 21A
, 21B. Next, the stored two types of image data Dinl and Din2 are converted into operand data DI and D.
2 is selected by the CPU 24 and outputted to the data calculation circuit 22.

The processing up to now is similar to the conventional example, but the present invention is different in the following points.

FIGS. 9(a) to 9(c) are explanatory diagrams of differential processing according to the first embodiment of the present invention, and FIG. 9(a) shows −th order differential processing of brightness image signal i (x). It shows the waveform.

In the same [type a), the vertical axis is brightness first-order differential data serving as image calculation data d1, which is obtained by performing negative-order differential processing on the signal waveform (x). The horizontal axis indicates the position X of the semiconductor component 29A. For convenience of explanation, the signal waveform will be shown as an analog waveform even after analog/digital processing, and the -order differential data d1 of the brightness image signal i (x) will be expressed as [1(x))'. shall be taken as a thing.

Moreover, the brightness image signal i (x) is
The differential processing is performed by the following, and the -order differential data d1 is individually stored in the first-order differential memory 22B. Here, the signal waveform peak indicates the position where the brightness of the semiconductor component 29A or the lead wiring 29C changes.

FIG. 6(b) shows a −th order differential processed waveform of the height image signal h(x).

In the same [ffl (b), the vertical axis is image calculation data d
2, which is the first-order height differential data, and the signal waveform h(x)
This is the result of −th order differential processing of . The horizontal axis is semiconductor component 2
9A position X is shown. Further, (h(x))' indicates the −th order differential data d2 of the height image signal h (x). Note that the height image signal h (x) is subjected to differentiation processing by the differentiation circuit 22A, and the -order differential data d2 thereof is individually stored in the first-order differentiation memory 22C. Here, the signal waveform peak indicates a position where the shape of the semiconductor component 29A or the lead wiring 29C changes.

FIG. 4(c) shows a quadratic differential processed waveform of the height image signal h(x).

In FIG. 3(c), the vertical axis is height second-order differential data serving as image calculation data d3, which is obtained by performing second-order differential processing on the signal waveform (h(x))'.

The horizontal axis indicates the position χ of the semiconductor component 29A. Also,
(h(χ))]2 is for displaying the second-order differential data d3 of the height image signal (h(X))'.The second-order differential data d3 is stored in the second-order differential data memory 22D. .

Here, the negative peak of the signal waveform indicates the position where the shape of the semiconductor component 29A or the lead wiring 29C changes to fall, and the positive peak thereof indicates the position where the shape changes to the rise.

- of the brightness image signal 1(x) differentially processed in this way
The second-order differential data d1 and the second-order differential data d3 of the height image signal [h(χ)]l are read out to the multiplication circuit 22H via the CPU 24.

FIG. 10 is an explanatory diagram of each edge detection circuit according to the first embodiment of the present invention, and shows the relationship between the multiplied composite waveform and the cross-sectional view of the printed circuit board.

In the figure, d4 is image calculation data, and signal i
Multiply value dlXd3 of negative differential data d1 of (x) and second differential data d3 of signal h (x) = (i(X) )'
This image calculation data d4 is calculated by the multiplication circuit 22E and output to the rising edge detection circuit 23A and the falling edge detection circuit 23B.

Here, P is the negative peak of the composite waveform, and the lead wire 2
This shows the tip position of 9C. This negative peak P is detected by the falling edge detection circuit 23B and outputted to the determination circuit 2 as the falling position P(X1) of the lead wire 29C.

In addition, q is the positive peak of the composite waveform, and the solder joint 29
This shows the protruding position of B. Similarly, the positive peak q is detected by the rising edge detection circuit 23A and output to the first determination circuit 28 as the protrusion position IP(x2) of the solder joint 29B.

In the first determination circuit 28, each detection circuit 23A and 23B
The first . Second image processing data Dout
1. Dout 2 and first reference data DRI serving as a determination standard are input, and first determination result data DAI is output. The determination process at this time is, for example, lead arrangement M2.
This is done by comparing the actual #lock distance l from the tip position of 9B with its separation reference value. Thereby, it is possible to inspect and determine whether the solder joint 29B protrudes from the tip of the lead wiring 29C of the semiconductor component 29A of the printed circuit board 29.

In this manner, according to the inspection device according to the embodiment of the present invention, the three-dimensional shape acquisition device 27 and the first determination circuit 28 are provided in the processing device.

Therefore, as in the conventional example, the actual separation distance from the tip position of the lead wiring 29C of the semiconductor component 29A of the printed circuit board 29! By image processing that compares the distance and its distance reference value, even when detecting the position P (x1) P (x2), the brightness image signal i (x) and the height image signal h (x ) It is no longer necessary to uniquely set the slice level SL for either one of the binary data.

For example, analog/digital processed signal i (x)
, h(x) is subjected to negative-order differential processing by the data calculation circuit 22, and then functional calculation processing such as multiplication processing is performed on the data subjected to the negative-order differential processing, thereby converting the negative peak p of the waveform to the semiconductor Lead wiring 29C of component 29A
Detecting =P(x1) of the tip position of the solder joint 29B from its positive peak q
) can be calculated with high precision. With this,
Actual separation distance at the tip position! The first determination result of high confidence orbit is obtained by comparing the first image processing data Dout in which the interference of errors in the detected amount is suppressed as much as possible with the first reference data DRI set in advance. Data DAI
It becomes possible to output.

This makes it possible to accurately inspect the mounting position of the object 29 to be inspected, such as a semiconductor component.

(ii) Description of the second embodiment 11. FIG. 12 shows the pattern g! according to the second embodiment of the present invention. FIG. 11 is an explanatory diagram of a recognition device, and FIG. 11 shows a recognition device incorporating an image processing device according to a second embodiment of the present invention.

In the figure, a device for non-contact recognition of the printing presence or absence of a character pattern 40 serving as the object to be recognized 2o includes at least the image processing device and the three-dimensional shape acquisition device 27. Second determination circuit 38. Stage drive device 30, height dictionary memory 31c
, a brightness dictionary memory 31D, and a data input circuit 36 are provided.

That is, 27 is a three-dimensional shape acquisition device which is an embodiment of the pattern image acquisition means 17, and is also used as the device used in the first embodiment. The acquisition device 27 acquires character pattern 4
0 images, for example, 1st. Second analog image signal A1n1. Brightness image signal 1(x
), which outputs a height image signal h (x).

31C and 31D indicate a height dictionary memory and a brightness dictionary memory that store height dictionary data D3 and brightness dictionary data D4 of the character pattern 40.

A data correlation calculation circuit 32 is an embodiment of the data calculation means 12, and includes a height data correlation circuit 32A, a brightness data correlation circuit 32B, and a multiple correlation calculation circuit 32C. The height data correlation circuit 32A uses the analog/digital processed height image signal h(x) and the height dictionary data D3.
It calculates the correlation between Further, the brightness data correlation circuit 32B calculates the correlation between the brightness image signal i (X) subjected to analog/digital processing and the brightness dictionary data D4. Further, the multiple correlation calculation circuit 32C calculates the correlation calculation data d11. which is an example of the calculation image data di, d2. d12 is combined (multiprocessed) and outputted to the position/matching degree detection circuit 33 as multiple calculation data d13.

A position/matching degree detection circuit 33 is an embodiment of the signal processing circuit 23, and detects the degree of multiple matching and the matching position of the character pattern 40, which is the content of the multiple calculation data d13.

Reference numeral 38 denotes a second judgment circuit which is an example of the second judgment means 1st, and which detects the third image processing data Dout3 outputted from the position/matching degree detection circuit 33 and the second criterion which is the judgment standard. It inputs data DR2 and outputs second determination result data DA2. The second reference data DR2 at this time is tolerance value data such as matching positional deviation of the character pattern 40.

Note that 39 is a design data file in which design data such as height dictionary data and brightness dictionary data of the character pattern 40 is stored.

A data input circuit 36 inputs and processes both data. Further, 30 is a stage driving device, which moves and scans a stage on which a printed circuit board or the like on which a character pattern 40 is printed is mounted.

Other components having the same names as those in the first embodiment have the same functions, so description thereof will be omitted.

Next, the operation of the recognition device will be explained.

FIGS. 12(a) to 12(e) are explanatory diagrams of the operation of the pattern recognition device according to the second embodiment of the present invention, and FIG. 12(a) shows an example of the pattern of the object to be recognized 20. .

In the figure, it is assumed that, for example, there is a request to recognize characters, shapes, etc. printed on a printed circuit board serving as the object to be recognized 20. Here, 40 is a character pattern, which is the Roman letter "A" printed on the recognition area A1 of the printed circuit board etc.
It shows. Further, 41 is the height shape pattern, which is a rectangular shape arranged in the recognition area A2.

FIG. 2B shows an example of a dictionary pattern of the object to be recognized 20. FIG.

In the figure, 42 is a dictionary character pattern, which indicates the contents of the brightness dictionary data D4 stored in the brightness dictionary memory 31D. Further, 43 is a dictionary shape pattern, which indicates the contents of the height dictionary data D3 stored in the height dictionary memory 31C. These data D3.
D4 is read from the design data file 39 and stored in each memory 31C, 31D via the data input circuit 36.

On the other hand, the image of the object to be recognized 20 is captured by the three-dimensional image acquisition device 27.
is obtained by the first . Second analog image signal A1n1. A1n2 is input to the image input circuit 35. At this time, the analog image signal A1n1. A1n2
represents the brightness image signal 1(x) and height image signal h(x) of the character pattern 40 and the shape pattern 41.

Similar to the first embodiment, these signals i (x) and h (x) are subjected to analog/digital processing in the image input circuit 35 and are individually stored as two types of image data Dint and Din2 in the height image memory. 31A, and stored in the brightness image memory 31B.

Furthermore, the image data Dinl and the height dictionary data D3 are C
It is selected by the PU 24 and inputted to the height data correlation circuit 32A as the operand data DI D3. Here, in the same figure (C), the vertical axis represents the degree of height matching, and the horizontal axis represents the relative position between the image data and the dictionary data. Further, the height matching degree is determined by comparing the image data Dinl and the height dictionary data D3 in the height data correlation circuit 32A, and determining the correlation with respect to the height and position. As a result, the shape pattern 41 is recognized.

Further, the image data Din2 and the brightness dictionary data D4 are selected by the CPU 24, and the operand data D2. D4
is input to the brightness data correlation circuit 32B. Here, in FIG. 3D, the vertical axis represents the degree of brightness matching, and the horizontal axis represents the position. Further, the brightness matching degree is determined by comparing the image data Din2 and the brightness dictionary data D4 in the brightness data correlation circuit 32B, and determining the correlation with respect to the brightness and position.

As a result, the character pattern 40 is recognized.

Thereafter, the correlation calculation data dll, d12 output from each correlation circuit 32A, 32B is sent to the multiplex correlation calculation circuit 3.
Multiple processing is performed by 2C. Here, in FIG. 4(e), the vertical axis represents the degree of multiple matching, and the horizontal axis represents the position. Further, the multiple matching degree is a result of multiple processing of the brightness matching degree and the height matching degree. The multiple calculation data d13 subjected to multiple processing at this time is
It is output to the matching degree detection circuit 33.

In position/matching degree detection step #I33, the multiple matching degree and matching position of the character pattern 40 are detected,
The third image processing data Dou output from the circuit 33
t3 is compared with second reference data DR2, which serves as a determination criterion. As a result, second determination result data DA2 is output from the second determination circuit. As a result, it is possible to determine whether or not a character pattern 402 as designed, such as the Roman letter 'AJ' or a shape pattern, exists in the recognition area A1 or A2, or whether or not the character pattern 402, such as the Roman letter 'AJ' or a shape pattern, exists in the recognition area A1 or A2, and whether or not the character pattern 402 exists, but is within the matching position deviation tolerance range. is done.

In this way, according to the recognition device according to the embodiment of the present invention, the processing device is provided with the three-dimensional image acquisition device 27, the second determination means 38, and the like.

Therefore, even if pattern recognition is performed by image processing the brightness image data and height image data of the character pattern 40, two or more types of operand data D
Slice level SL for any one of data 1 to D4
For example, the data correlation calculation circuit 32 performs correlation processing on four types of operand data D1 to D4 without uniquely setting processing, and then performs functional calculation processing such as multiple processing on the correlated data. By doing this, it is possible to accurately recognize and process the positional shift of the character pattern 40 based on the sharpness of the matching degree. As a result, by performing a comparison process between the second image processing data Dout having a high degree of matching and the second reference data DR2 set in advance, the second determination result data DA2 having a high degree of accuracy is output. It becomes possible to do so.

This makes it possible to accurately recognize the mounting condition of the object 20 to be recognized, such as a semiconductor component.

In each of the embodiments of the present invention, two types of image data including the position coordinates of the object to be inspected 19 and the object to be recognized 20 are used, but the reflectance data, red, blue,
The present invention can also be applied to image processing using color data such as green.

〔Effect of the invention〕

As explained above, according to the processing device of the present invention, two or more types of image data are subjected to differential processing or composition processing, or
A data calculation means is provided which performs the differential processing, composition processing, etc. and outputs one or more types of image calculation data.

Therefore, even if image processing is performed to detect the distance H distance from a certain position of the image acquisition target to an arbitrary position as in the conventional example, the position coordinates of the object can be accurately determined from the waveform falling position. It becomes possible to perform calculations.

Moreover, according to the inspection apparatus of the present invention, the processing device is provided with an object image acquisition means and a first determination means.

Therefore, by performing functional calculation processing such as first-order differentiation processing and multiplication processing on two or more types of operand data by the data calculation means, the reliability of each data is improved, and the position coordinates of the object to be inspected can be determined with high precision. It is possible to perform arithmetic processing.

This makes it possible to accurately inspect the mounting position of the object to be inspected, etc., based on the first image processing data in which the intervention of errors is suppressed as much as possible.

Furthermore, according to the recognition device of the present invention, the processing device is provided with a pattern image acquisition means and a second determination means.

Therefore, two or more types of operand data are subjected to correlation processing by the data calculation means, and then subjected to multiple processing. This makes it possible to accurately recognize the mounting state of the object to be recognized, such as a semiconductor component, based on the second image processing data with a high degree of matching.

This greatly contributes to the manufacture of highly reliable and highly accurate image processing devices, object inspection devices, and pattern recognition devices.

[Brief explanation of drawings]

FIG. 1 is a principle diagram of an image processing device according to the present invention, FIG. 2 is a principle diagram of an object inspection device according to the present invention, FIG. 3 is a principle diagram of a pattern recognition device according to the present invention, and FIG. 5 is a diagram illustrating the operation of the image processing device according to the first embodiment of the present invention. FIG. 6 is a diagram illustrating the operation of the image processing device according to the first embodiment of the present invention. FIG. 7 is an explanatory diagram of an object to be inspected according to the first embodiment of the present invention; FIG. 8 is a diagram of the object inspection apparatus according to the first embodiment of the present invention; FIG. 9 is an explanatory diagram of the differential processing according to the first embodiment of the present invention. FIG. 10 is an explanatory diagram of each engine detection circuit according to the first embodiment of the present invention. FIG. 11 is a block diagram of a pattern recognition device incorporating the image processing device according to the first embodiment of the present invention, and FIG.
FIG. 13 is an explanatory diagram of the operation of the pattern recognition device according to the second embodiment of the present invention, and FIG.
FIG. 4 is a diagram illustrating problems related to the conventional example. (Explanation of symbols) 11... Storage means, 12... Data calculation means, 13... Signal processing means, 14... Control means, 15... Object image acquisition means, 16.18... 1st. 2nd determination means, 17... Pattern image acquisition means, 18... Optical path division means, M1-Mn... 1st - nth storage means, Dinl-D
inn-first to nth image data, D1 to Dn...first to nth operand data, d1 to dn...first to nth
Image calculation data of Dout Dout 1,
Dout 2-Image processing data. 1st. Second image processing data, DPI, DR2...first. Second reference data, DA
I, DA2... 1st. Second judgment result data. 〆11 Memory means

Claims (5)

[Claims] (1) Store n types of image data (Din1 to Dinn) and store at least two or more types of operand data (D1 to Dinn).
=Dini, Dn=Dinj).
1) and the two or more types of operand data (D1 to Dn)
is subjected to functional calculation processing to obtain one or more types of image calculation data (d
data calculation means (12) for outputting image calculation data (d1 to dn); and signal processing means (13) for signal processing the image calculation data (d1 to dn) and outputting image processing data (Dout).
and a control means (
14) An image processing device comprising: (2) The image processing apparatus according to claim 1, wherein the storage means (11) stores first to nth image data (Din1 to
The data calculation means (12) consists of first to nth storage means (M1 to Mn) that individually store the data (Dinn) and individually output the first to nth operand data (D1 to Dn).
is at least the first to nth image data (Din
1 to Dinn), or performs differential processing and synthesis processing on two or more types of operand data (D1 to Dn). (3) An object inspection device comprising the image processing device according to claim 1, wherein the image processing device includes at least an object to be inspected (19).
images are acquired and the n types of image data (Din1 to
The first image processing data (Dout1) outputted from the signal processing means (13) and the first reference data (DR1) serving as a determination standard are inputted to The first judgment result data (DA
1) A first determination means (16) for outputting a first determination means (16). (4) The object inspection apparatus according to claim 3, wherein the data calculation means (12) at least performs image data (Dini) selected from the first to n-th image data (Din1 to Dinn). ), and multiplies the data subjected to the n-th differential processing. (5) A pattern recognition device comprising the image processing device according to claim 1, wherein the image processing device includes at least a target to be recognized (20).
images are acquired and the n types of image data (Din1 to
a pattern image acquisition means (17) that outputs a pattern image (Dinn), second image processing data (Dout2) output from the signal processing means (13), and second reference data (DR2) that serves as a pattern judgment standard. Second determination means (18) for inputting and outputting second determination result data (DA2)
A pattern recognition device comprising: