JPH0120477B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a method and apparatus for inspecting patterns on printed matter.

[Problems to be solved by conventional technology and invention]

As shown in Fig. 1, the conventional printed matter pattern inspection device extracts image information while inspecting the printed matter 1 in the width direction with a line sensor camera 2, and on the other hand, uses the output of a rotary encoder 4 attached to a cylinder 3 of the conveyance system. By synchronizing the scanning of the camera 2, the sample pattern 5 (FIG. 2) is divided into a large number of small pixels arranged in a matrix.

The density of each of these pixels is sequentially compared with the corresponding small pixels of the sample pattern 6 (FIG. 2), which has been similarly divided and recorded in the memory, to detect the presence or absence of a defect.

In this kind of pattern inspection, if there is a positional shift in the width direction of the conveyance system as shown in Figure 3, the gradation obtained from the specimen will change for the 3 marked pixels out of the gradation 7 obtained from the sample. When compared with the corresponding three pixels of No. 8, it is erroneously determined that there is a defect even if there is no defect in the sample pattern.

One way to prevent this is to set a large tolerance value for the tone difference between the sample and the sample.

However, this results in a decrease in defect detection accuracy and is not necessarily a good countermeasure.

A method free from such problems involves correcting the positional deviation of the sample pattern with respect to the sample pattern, and various methods have been devised.

The first method is as shown in Japanese Patent Application No. 56-146355 filed by the applicant of the present invention, which detects the position of a pattern on a printed matter in the width direction, and detects the position change when writing sample pattern data into a reference value memory. When the data of the sample pattern and the data of the sample pattern are written in the detected value memory are different by a predetermined value or more, the sample pattern data at that time is written in the reference value memory.

However, this method requires a device for detecting positional changes in the width direction of the printed material and for correcting positional deviations according to the positional change information.
This results in problems of increased hardware complexity and higher costs.

Part 2 is shown in Special Publication No. 55-45948, etc.
This method compares a sample pixel with several pixels in the vicinity of the corresponding sample pixel.

However, this method has a problem in that it inevitably reduces inspection accuracy.

The present invention has been made in view of the above-mentioned points, and an object of the present invention is to provide a method and apparatus for inspecting patterns of printed matter, which can inspect patterns with high precision even if there is a positional shift in a printed matter conveyance system, and which can be constructed at low cost. With the goal.

[Means to solve the problem]

In order to achieve the above object, in the first invention of the present application, sample data detected by decomposing a sample pattern of a printed matter into a pixel matrix is compared pixel by pixel with sample data extracted in advance from a sample printed matter and stored in a memory. In a pattern inspection method for a printed matter for determining the quality of a pattern, one of the sample data and the specimen data is shifted in the horizontal direction of the printed material by one pixel and compared with the other of the two data, and both data are closest to matching. A pattern inspection method for a printed matter, characterized in that the sample data or the sample data is corrected for the positional shift in the horizontal direction of the printed matter based on the amount of positional shift between the two pieces of data when In a second invention of the present application, there is provided: a device for forming a synchronization signal according to the conveyance operation of a printed matter; a camera that scans a pattern of the printed matter pixel by pixel based on the synchronization signal and extracts image data; A specimen data memory in which image data extracted from specimen prints by this camera is written, and specimen data from the camera and data from the specimen data memory are shifted one pixel by one pixel from the other. A circuit detects the amount of positional deviation when both data come closest to matching, and a circuit that performs positional deviation correction based on the output of this positional deviation amount detection circuit, compares both data pixel by pixel, and determines whether the pattern of the sample print is good or not. To provide a pattern inspection device for a printed matter, which is equipped with a circuit for making a determination.

[Effect]

Extract sample data from the sample print. This sample data is compared with sample data previously stored in memory. This comparison is performed for both sample data and sample data, while moving one pixel at a time in the horizontal direction of the printed matter with respect to the other.
Through such a comparison, the amount of positional deviation can be determined from the amount of shift when both data are closest to each other. Then, after correcting the positional deviation using this positional deviation amount, the sample data and the sample data are compared to determine the quality of the pattern of the sample print.

〔Example〕

FIGS. 4a, b to 8 show an embodiment of the present invention.

Figures 4a and 4b show sample data and specimen data; Figure 4a shows the basic relationship between sample data and specimen data handled in the method of the present invention, and Figure 4b shows the pattern of printed matter. This shows the relationship between sample data in special parts such as edges.

First, according to figure a, a plurality of sample data 10 to 16 are referred to for one sample data 9. These sample data 10 to 16 are obtained by shifting the basic sample data 13 one pixel at a time to the left and right by the expected positional shift.

Then, sample data 9 is changed to sample data 10 to 16.
Among them, the density can be compared with the sample data 12 which is closest in phase to the sample data 9, and defect detection is thereby performed.

As a result, even if the sample data 9 is misaligned in the width direction of the printed material, the data can be compared in terms of content, so defects can be detected with high accuracy.

Next, according to FIG.
This has a large gradation difference tolerance. This makes it possible to carry out defect inspection suitable for cases where the gradation difference is extremely large, such as at the edge of a picture. That is,
Since the sample data 18 has the closest phase to the sample data 17, the phase difference between the two is 1/2 pixel width. Therefore, the tone difference between the sample and the specimen at the pixel corresponding to the edge when a positional shift of 1/2 pixel width occurs may be taken as the tolerance value for that pixel. Various methods can be used to set the permissible value for each pixel (see Japanese Patent Application No. 179599/1983), or pixels corresponding to edges may be masked to be excluded from the determination.

FIG. 5 shows the configuration of an apparatus for carrying out the method of the present invention, which is constructed based on the idea shown in FIG. In this configuration, the image on the printed matter 1 is scanned by the camera 2 and image information is extracted.
At this time, the rotation amount of the conveyance system cylinder 3 is taken out by the rotary encoder 4 and given to the synchronization circuit 21. The synchronization circuit 21 provides its output to the camera 2, the A/D conversion circuit 22, and the address circuit 24, and performs scan synchronization of the camera 2, A/D conversion synchronization of image information taken out by the camera 2, and sample data memory. 25 and pixel control data memory 2
6 address assignment is performed.

Based on this address assignment, the sample data S corresponding to each sample data T is read out from the sample data memory 25 into which sample pattern data has been taken in by the changeover switch SW at the start of the test, and the pixel control data is read out from the pixel control memory 26. Read M. The pixel control data M is data for giving a permissible value to each pixel.

Sample data S, pixel control data M and A/D conversion circuit 22 in these memories 25 and 26
The sample data T given through is input to the positional deviation correction circuit 23 and corrected as S', M', and S', respectively.
becomes T′. Then, the determination circuit 27 detects a pattern defect in the sample print based on the corrected data S', M', and T'.

Figure 6 shows the positional deviation correction circuit 23 in Figure 5.
This figure shows the configuration of a correlator that forms part of the correlator. This correlator is used to configure a positional deviation amount detection circuit (Figure 7), and based on this detection signal, a correction circuit (Figure 8) is constructed.
is configured to perform a positional deviation correction operation.

For example, a TDC1023J manufactured by TRW may be used as the correlator 30, and this correlator includes two 64-bit shift registers 31, 32, one 64-bit latch 33, and an exclusive NOR gate ENOR ₁ to
ENOR ₆₄ and an adder 34.

The shift register 32 is supplied with a 1-bit serial input B _IN (64 bits) of sample data and a clock CLKB for serial input, and each bit output is supplied to the latch 33 (R ₁ to R ₆₄ ). Signal LDR is applied to these latches and latched. The shift register 31 also has a 1-bit serial input A _IN (64 bits) of sample data and a clock CLKA for serial input.
are given, each bit output thereof becomes one input of exclusive NOR gates ENOR ₁ to ENOR ₆₄ , and each bit output of latch 33 (R ₁ to _{R 64} ) is given to the other input. Each output of exclusive NOR gate ENOR ₁ to _{ENOR 64} is added to adder 3.
The adder 34 is configured to output the number of exclusive NOR gate outputs, that is, the correlation value (Corr) in 7 bits each time the clock CLKS is applied.

In this correlator, first, 64-bit sample data _B1 ... _B64 is input to shift register B by clock CLKB, and then the contents of each bit of shift register 32 are changed to latch 33 by latch signal LDR. R ₁ to R ₆₄ ). A correlation value is then calculated between each output of the shift register 31 and the latch 33, and during this calculation time the shift register 3
2, the following sample data can be input.

Sample data held in latch 33 R ₁ ...R ₆₄
An exclusive NOR is taken for each corresponding bit between the sample data _A1 ... _A64 inputted to the shift register 31 by the clock CLKA, and
When A _o and R _o match, ENOR _o = 1, and when they do not match, ENOR _o = 0 is obtained. Of these, ENOR _o
The number of bits for which =1 is output by the adder 34 as a correlation value Corr. The maximum value of the correlation value Corr is 64.

The correlation value Corr takes a different value each time sample data A is shifted, but sample data A and sample data R
Matching the phases with the above is equivalent to the correlation value Corr taking the maximum value, and the amount of positional deviation can be detected by checking the phase at which the correlation value Corr takes the maximum value.

FIG. 7 shows a positional deviation amount detection circuit constructed using the correlator 30 shown in FIG.

First, to give a conceptual explanation, this circuit outputs the amount of positional deviation D, and D takes a value from -8 to +7. If D<0, the phase of the sample data T is advanced; If D=0, it represents a state in which the phases of the sample data T and sample data S are equal, and if D>0, it represents a state in which the phase of the sample data S is advanced. In other words, the state in which sample data T is ahead by 8 bits is D.
=-8, and the state in which the sample data T is delayed by 7 bits corresponds to D=+7. The detectable positional deviation width can be easily expanded as described later.

In order to compare the phases of sample data S and sample data T, the correlator 30 takes the correlation between the time series 1/0 patterns of MSBS ₇ and T ₇ of each data. Here, both the sample data S and the sample data T are 8-bit data that takes values from 0 to +255. Therefore, when the sample data S (sample data T) has a value of +128 or more, S ₇ =1 (T ₇ =1), and when it is less than 128, S ₇ =0. That is, this corresponds to setting gradation 128 as a threshold, and assigning 1 to pixels with a gradation equal to or higher than the threshold, and φ to pixels with a gradation less than the threshold.

In this method, both data S, T
If all the gradations are 128 or more or all less than 128, there will be no correlation between the two data S and T, and it will be impossible to detect the amount of positional deviation. However, in general pictures, there are various gradations from dark to bright levels. Since the above-mentioned cases are rare, the method described here has a great effect even though it has the simplest hardware configuration.

If there is a problem with such a detection method, the following method may be adopted.

(i) The threshold value is not fixed at 128, but is given floatingly according to both data S and T. For example, a method may be considered in which the average value of both data S and T over one scan is set as a threshold value.

(ii) Find the edges of the picture by differentiating both data S and T, and use the time series 1/0 patterns S' and T' corresponding to the edges in place of S ₇ and T ₇ to calculate S', Correlate T′.

Next, the specific operation of the circuit shown in FIG. 7 will be explained.
Suppose that sample data S and sample data T that have the same address are input at the same time, and in order to detect the state in which the phase of sample data T is leading, the sample data
_T7 is input to an 8-bit shift register 41, and its phase is delayed by 8 bits. The scanning clock CLK of the camera may be used as the clock for shifting the shift register 41.

When the 64-bit time series 1/0 pattern _S7 to be a sample is input to the correlator 30, the counter 42 outputs a signal LDR that latches the pattern.
Latch 44 holds the correlation value Corr. Latch 4
The initial value at the start of camera scanning of No. 4 is 0, and is initialized by the scanning start signal S ZERO generated by the system. Then, each time a larger correlation value Corr is output from the correlator 30, the comparator 4
5 is updated to the larger value.

A counter 43 counts the scanning clock CLK of the camera, and a latch 46 holds the value (4 bits, value from 0 to 15). Since the value of the latch 46 is updated by the output of the comparator 45 every time a larger correlation value Corr is output, the latch 46 maintains the phase when the correlation value Corr reaches its maximum value. become. The subtractor 47 calculates "8" from the maximum correlation value Corr held in the latch 46.
By subtracting the amount of positional deviation D (-8 to +7)
can be obtained. The reason for subtracting "8" is to correct the delay caused by the shift register 41 by the amount by which the counter 43 counts eight clock pulses CLK.

The counter 43 and latch 44 store sample data.
Since the MSB _S7 is initialized every time 64 bits are input to the correlator 30, the amount of positional deviation D between the sample data S and the sample data T is output for every 64 pixels. In other words, the total number of pixels of the line sensor camera is
512, it is possible to correct the positional deviation between the sample data S and the sample data T for each of eight blocks divided into 64 pixels each.

If you want to detect the value of the positional deviation amount D (-8 to +7) over a wider range, for example, change the shift register 45 to 16 bits, change the counter 43 to 8 bits output, and use the subtracter 47 to output "16". By reducing the positional deviation amount D (-16 to +15)
get. Thereafter, a wide range of positional deviation amounts can be detected in the same manner.

Figures 8a and 8b show circuits that perform positional deviation correction based on the output D of the positional deviation amount detection circuit in Figure 7. Figure 8a is a circuit that handles sample data T, and Figure 8b is a circuit that handles sample data T. Each circuit that handles data S is shown, and both circuits have substantially the same configuration. That is, 8-bit shift register 5
The positional deviation correction value T _o ' of the n-th bit T _o of the sample data T is obtained by the 1 to 1 and 16 to 1 selector 52 . If this configuration is made 8-bit parallel, data T' obtained by correcting the positional deviation of the sample data T can be obtained. Here, when D≧0, T′=t ₀ (t ₀ : no shift), D<
When it is 0, T'=t ₁ , . . . t ₈ (shift of 1 bit to 8 bits).

Similarly, when D≦0, S'= _s0 , M'=
m ₀ , when D>0, S′=s ₁ ~s ₈ , M′=m ₁ ~ _{m 8} (1
bit to 8 bit shift). The shift registers 51, 53, and 55 are initialized to 0 by a scan start signal S ZERO generated by the system at the start of camera scan.

As described above, sample data S', sample data T' and pixel control data M' whose positional deviations have been corrected according to the positional deviation amount D are obtained. These corrected S′,
T' and M' are input to the determination circuit 27 (FIG. 5) and defects are detected.

FIG. 9 shows an embodiment of the determination circuit 27, with 57 to 6
2 is an 8-bit data latch circuit, 63 is an OR gate array configured to allow 8-bit input, 64 is a shift register, and 65 is an 8-bit data latch circuit.
66 and 69 are adder circuits, 67 is an inverter, 68 is an exclusive OR gate, and 70 is a comparator.

Next, the operation of this embodiment will be explained.

When the printing operation starts and the inspection operation begins, the sample data S′, sample data T′, which have been corrected for positional deviation, are
Pixel control data M' is sequentially loaded into each latch 57-60.

Therefore, among these, the positive state sample data S' input to the latch 59 and the negative state sample data T' input to the latch 60 are transferred to the adder circuit 66, the inverter 67, and the exclusive OR gate 6.
8. The data D is processed by a circuit consisting of an adder circuit 69 to become data D representing the absolute value of the difference between the sample data T' and the sample data S', and is written into the latch 62. That is, D
= |T′−S′| is extracted. This data D is then supplied from latch 62 to one input B of comparator 70.

In addition, the reference data taken out from the latch 58
S' is input to the shift register 64, where it is shifted by a predetermined number of bits and multiplied by a predetermined coefficient K (less than 1) to become data KS', which is then supplied to one input B of the selection circuit 65. Ru.

On the other hand, the pixel control data taken into the latch 57
M' is read out as is and supplied to the other input A of the selection circuit 65, and is also supplied to the selection input S of the selection circuit 65 via the OR gate 63. The selection circuit 65 functions to output the data supplied to the input B when the selection input S is at the "H" level, and outputs the data at the input A when the selection input S is at the "L" level.

Comparator 70 compares data JL of input A with data D of input B, and generates an output J that becomes "H" level only when the data of input B is greater than the data of input A. That is, a comparison operation is performed using the data JL as the determination level, and the output J is obtained.

With the above configuration, even if the pattern on the printed matter is misaligned to some extent, it is possible to accurately determine whether the pattern is good or bad, enabling high-speed inspection and reducing the cost of the apparatus. Furthermore, since the sample data is compared with the sample data, it is possible to perform an inspection with much higher precision than when comparing the patterns as a whole.

In the embodiments described above, the sample data may not be stored in the memory, or a sample data memory may be provided. Further, when detecting positional deviation, the sample data may be displaced one pixel at a time in the horizontal direction of the printed material.

〔Effect of the invention〕

As described above, when comparing the sample data extracted from the sample print with the sample data stored in the memory, the present invention shifts one of the sample data and the sample data from the other one pixel at a time in the horizontal direction of the print. The amount of positional deviation is determined based on the amount of shift where both data come closest to matching, and the positional deviation is corrected based on this amount of positional deviation, and then the sample data is compared with the sample data to determine the pattern of the sample print. Since the inspection is carried out, even if there is a positional shift in the print conveyance system, the pattern inspection can be carried out with high precision, and the apparatus configuration can be done at low cost.

[Brief explanation of drawings]

FIG. 1 is a diagram showing the configuration of an image information detection section of a pattern inspection device that is the object of the present invention, FIG. Fig. 3 is an explanatory characteristic diagram when a positional shift of a pattern occurs due to the print conveyance system, etc., and Fig. 4 a and b are characteristics showing the basic inspection method of the method of the present invention and the pattern edge inspection method. figure,
FIG. 5 is a block diagram showing the configuration of a device for implementing the method of the present invention, FIG. 6 is a block diagram showing the internal configuration of a correlator used to configure the device of the present invention, and FIG. 8 is a connection diagram showing the configuration of a positional deviation detection circuit configured using the correlator shown in the figure; FIG. 8 is a block diagram showing the internal configuration of a circuit that is combined with the circuit of FIG. FIG. 9 is a block diagram showing a more specific configuration of the determination circuit in the apparatus of FIG. 5. 1... Printed matter, 2... Camera, 3... Cylinder, 4... Rotary encoder, 5, 7, 9,
17...Sample picture (sample row) data, 6, 8, 1
0,18...Sample picture (sample row) data, 31,
32, 41, 51, 53, 55...shift register, T...sample data, S...sample data, M...
...Pixel control data, JC...Judgment circuit, L...Latch, SR...Shift register.

Claims (1)

[Claims] 1. Sample data detected by decomposing a sample pattern of a printed matter into a pixel matrix is compared pixel by pixel with sample data previously extracted from the sample printed material and recorded in a memory to determine the quality of the pattern. In a pattern inspection method for a printed matter, one of the sample data and the sample data is shifted in position in the horizontal direction of the printed material by one pixel and compared with the other of the two data, and when the two data come closest to matching, the two data are detected. A pattern inspection method for a printed matter, characterized in that the sample data or sample data is corrected for positional deviation in the horizontal direction of the printed matter based on the amount of positional deviation between them, and then compared with the sample data. 2. A device that generates a synchronization signal according to the conveyance operation of the printed material, a camera that scans the pattern of the printed material pixel by pixel based on the synchronized signal and extracts image data, and a device that generates image data extracted from the specimen printed material by this camera. The specimen data memory to be written, the specimen data from the camera, and the data from the specimen data memory are compared while shifting one of the data from the other in the horizontal direction of the print by one pixel, and both data are the best. It is equipped with a circuit that detects the amount of positional deviation that approaches a match, and a circuit that performs positional deviation correction based on the output of this positional deviation amount detection circuit, compares both data pixel by pixel, and determines the quality of the pattern of the sample print. Pattern inspection device for printed matter.