JPH04152479A - Image inspecting system - Google Patents

Abstract

PURPOSE:To perform the inspection without using a standard pattern by extracting a signal caused by an image failure from a reading signal, and executing the inspection, based on a distance extending from a picture element corresponding to the extracted signal to the vicinity of a boundary of the image. CONSTITUTION:An output of an image input device 11 is supplied to a threshold processing circuit 13, and the circuit 15 executes binarization by using two thresholds set by a CPU 15, and outputs a signal corresponding to an outline and a defect of an image. An output of the circuit 13 is stored in a binary image memory 19 under the control of a memory control circuit 17, and the CPU 15 converts a binary image to a distance image, and detects a picture element of a printing blur from the distance image. The image signal converted to the distance image is stored in a distance image memory 23, and information related to the detected picture element of a printing blur is offered to a deciding output circuit 25, and outputted to the outside. Accordingly, according to this inspecting system, an inspection of an image of printing, etc., can be executed satisfactorily without preparing a standard pattern, and also, an area in which a defect of the image is generated can be detected in detail.

Description

DETAILED DESCRIPTION OF THE INVENTION [Object of the Invention] (Industrial Application Field) The present invention relates to an image inspection method, and particularly to an image inspection method for inspecting printed matter for defects.

(Prior Art) An image inspection method, particularly a technique for inspecting defects (fading, etc.) in printed matter, is described in Japanese Patent Publication No. 1-20477. According to the technology described in this publication, sample data of a printed matter is compared with a standard pattern to determine the quality of the printed matter. However, although such technology can certainly determine the quality of printed matter, it is not possible to know the cause of printing defects, and it is necessary to create a standard pattern for each inspection target.

On the other hand, IEICE Spring National Conference Proceedings D-268 "Evaluation algorithm for image unevenness" (Benjo et al.).

A technique for examining the statistics of density information of an image is also known, as shown in (1989). This technique has the disadvantage that it is not possible to detect in detail the area and cause of the defect.

(Problems to be Solved by the Invention) As described above, the conventional techniques have the drawbacks that a standard pattern is required and that a region where a defect occurs cannot be detected in detail.

SUMMARY OF THE INVENTION In view of the above drawbacks, it is an object of the present invention to provide an image inspection method capable of inspecting images without using a standard pattern.

A further object of the present invention is to provide an image inspection method that can detect in detail areas where image defects have occurred.

[Structure of the Invention] (Means for Solving the Problem) In order to achieve the above object, the image inspection method of the first invention includes a reading means. The reading means reads the surface of the object to be inspected containing the image and converts it into a corresponding signal. From the signals converted by this means, signals including signals caused by the image defects are separated and extracted. The distance from the pixel corresponding to this separated signal to the vicinity of the outline or boundary of the image (boundary between printing and non-printing areas) is determined. The present invention is characterized in that the image is inspected based on the distance obtained by this means.

The second aspect of the invention separates and extracts signals including signals caused by image defects from the read signals.

From this signal(s), signals corresponding to the contours (or boundaries) of the image are deleted. After this deletion, the image is inspected from the residual signal.

(Operation) The signals obtained by the reading means include those corresponding to one image on the object to be read. In particular, when the image includes defects such as blurring or cuts, information corresponding to the defects is also included in the signal. For example, the information corresponding to the defect is that the defect is abnormal in terms of density and that the position of occurrence is near the contour or boundary of the image (boundary between printed and non-printed areas).

Therefore, abnormal signals (including defective signals) are extracted from the read signal. Then, in the first invention, how far these signals are separated from the boundary is checked. Since defective signals exist far from the boundary, they can be distinguished from normal signals.

Further, in the second aspect of the invention, since the lower signal corresponding to the boundary is deleted from the abnormal signal, the remaining signal becomes the signal corresponding to the defective pixel.

(Example) Embodiments of the present invention will be described in detail below with reference to the drawings.

FIG. 1 is a diagram showing the configuration of an image inspection apparatus according to this embodiment. As shown in the figure, this image inspection device
It includes an image input device 11. This device 11 is, for example, a scanner, and includes a light source and a photoelectric conversion section that receives light emitted from the light source and reflected from the image to be read, and converts it into a corresponding electrical signal, although the components thereof are not shown. Consisting of

If necessary, the image input device 11 may be a color image input device capable of reading color images.

The output of this image input device 11 is an image signal holding density information, and is, for example, an 8-bit signal. This signal is supplied to the threshold processing circuit 13. This circuit 1
In step 3, binarization is executed using two threshold values set by the CPU 15. However, rather than simple binarization, r
Give lJ, and give "0" to everything else.

On the other hand, the output from the threshold processing circuit 13 is a 1-bit signal, which is stored in the first binary image memory 19 under the control of the memory control circuit 17.

Note that this memory control circuit 17 receives a signal indicating the start of image reading processing from the image input device 11 and controls the storage operation in the first binary image memory 19 described above. When this storage operation is completed, the memory control circuit 17 outputs a signal indicating completion to the CPU 15 L1, after which the first binary image memory 19 comes under the control of the CPU 15 and is used for signal processing.

Signal processing executed by the CPU 15 is performed according to a program stored in the program memory 21. This C
The processing in the PU 15 includes a distance image conversion process from a binary image to a distance image, as will be described later. From the distance image obtained through this conversion process, a process of detecting pixels with blurred printing (hereinafter referred to as blurring detection processing) is performed. Note that during this blurring process, the second binary image memory 22 is used.

The image signal converted into a distance image is stored in the distance image memory 23.
is memorized. Information regarding the detected pixels with blurred printing is provided to the determination output circuit 25 and output to the outside. During the above processing, the CPU 5 uses the working memory 27
Execute the process using . Note that the components related to the processing in the CPU 5 (in this embodiment, the CPU 15
, first 21 image memory 19, program memory 21,
The distance image memory 23, the judgment output circuit 25, and the working memory 27 are connected by a data bus 29 and an address bus 31.

Next, the processing by the CPU 15 will be explained.

As described above, the processing performed by the CPU 15 is the distance image conversion processing and the blur detection processing, and as preliminary processing, the threshold processing circuit 13 performs processing for obtaining two threshold values. These two threshold values are, for example,
There is a method of determining the density occurrence opening degree of the image signal obtained in . In this method, a density histogram is obtained and two threshold values are set from this density histogram.

In this example, density refers to a physical quantity that is proportional to the amount of light reflected from the object to be read, where a low density means closer to black, and a higher density means closer to white. It is shown that. Further, in this embodiment, the image signal representing density is expressed in 8 bits as described above.

The method of setting the threshold value here will be explained using FIG. 2. Now, the density histogram is expressed by h(x). This shows that there are h (x) things of concentration X. Note that X has a value of 0 to 255. A smoothing process is performed on this density histogram h (x). This is for noise removal. This smoothing process is realized by simple averaging, and if the density histogram after the smoothing process is ha(X), ha (x)= (h (x-1) +2h (X)+
It is expressed as h (x+1) ) /4.

For this density histogram ha (x), two areas Ra and Rb. Density values (hereinafter referred to as pealing densities PP and PB) that give the maximum frequency value within the respective regions Ra and Rb are detected. Here, the area Ra is a low-density signal area and corresponds to the image area (printed area). This area Ra is hereinafter referred to as a print area. Further, the region Rb corresponds to the background portion (referring to the region other than the above-mentioned image portion). Here, this area R
b is called the background area.

Next, standard deviations σ and σ of the density frequencies of the print area Ra and background area Rb are determined from the density histogram.

Then, σ, −Σ (h(i)−＾VEa) ” /Σ 1−
1 R- a b -X (h (i) AVEb
) 2/ Σ However, ＾VEa-Σ h(t)/Σ 1 AVEb-Σh (i) /E 1.

Such peak concentration values PP and PB and standard deviation σ
, and σ, two threshold values TL and TI are determined as follows.

TL-PP+A*σ.

TH-PB-B*σ.

Here, the constants A and B are determined experimentally, and the inventor's experiments have shown that a range of 3 to 5 is preferable. As above, two threshold values TL
, T)l is found. Other methods for determining these two thresholds will be described later.

Next, the actual process will be explained based on FIG. 3. First, as the image input device 11 starts operating, a binary image is stored in the binary image memory 19 under the control of the memory control circuit 17.
(step 1). As described above, the binarization in the threshold circuit 13 gives "1" to the signal with a concentration existing between the two thresholds TL and T)I, and gives "0" to the other signals. give. For example, when a printed number "0" as shown in FIG. 14(A) is read along the dotted line, a signal such as curve C in FIG. 14(B) is obtained. In the same figure (B), the horizontal axis is the position, and the vertical axis is the density. Threshold values TL and TI are set for this signal as shown. Therefore, the signal output from the threshold circuit 13 and stored in the first binary image memory 19 corresponds to the signal between the threshold values TL and TH. Therefore, as can be seen from FIG. 3, the output of the threshold value circuit 13 consists of one that corresponds to the outline of the image and one that corresponds to (what appears to be) a defect in the image.

Next, the binary image obtained in step 1 is converted into a distance image (step 2). The process in step 2 will be described in detail later, but each point of the binary image is expressed by the distance from the boundary (contour) of the original image.

After conversion to a distance image, the number of pixels having a distance value greater than or equal to a predetermined distance is counted (step 3). From this counting result, the number of defective pixels located within a predetermined distance or more from the boundary of the original image can be determined. Therefore, print quality is determined from this result (step 4). This completes the entire inspection process.

Next, regarding the conversion process to a distance image in step 2, the fifth
This will be explained according to the diagram.

First, an overview of the processing will be explained. As described above, the binary image to be processed (stored in the first binary image memory 19) consists of signals corresponding to the contours of the image and signals corresponding to defects in the image. Among these signals, the signal corresponding to the outline of the image disappears as soon as the "image" is narrowed.

On the other hand, defects in the image do not disappear immediately even if the image is narrowed down. This is because image defects usually occur near the outline of the image. Here, outline means the boundary between the printing area and the non-printing area. Focusing on such characteristics, in the following processing, the image narrowing process is repeated, and the number of times the narrowing process has been performed is stored for the remaining pixels, and this number is used as a distance image value.

The above processing will be explained in detail. The 1% Js distance image d(i, j) is used as a variable in this process. However, 0≦i<WIDTHO≦
j < HEIGHT. Here, W
IDTH represents the width of the image read by the image input device 11, and HEIGHT represents the height of the image. This information is obtained from the image input device 11.

First, as step 2-1, the above distance image d(i,
Initialize j). That is, d (i, j) −
Set as 〇.

Next, the variable "narrowing processing number" n is set to 1 (step 2-2). Next, the distance conversion operation flag f is set to 0 (step 2-3). In order to identify the pixel to be subjected to the conversion process, we introduce the image vertical position y as a variable, and y−
1 (step 2-4). Similarly, the image horizontal position X is introduced as a variable and set to X-1 (step 2-5). After these settings, the first binary image memory 19
It is checked whether the value of the image stored in is 1 (step 2-6). Specifically, the first binary image memory 1
Among the pixels stored in 9, the variable (x.

' Recall the value of the pixel specified by (1, 1). Check whether the value of this called pixel is 1 or not. As a result of the check, if the value of the pixel stored in the first binary image memory 19 is not 1, the converted image g (x, y
) is set to 0 (step 2-7). In this case, since the value of the pixel stored in the first binary image memory 19 is 0, there is no defect or the like, so the converted image is also 0.

After this step 2-7, the process moves on to checking the next pixel. First, add 1 to variable X (step 2-8),
Check whether variable X is smaller than (WIDTH-1) (step 2-9).

If the variable X is smaller than (WIDTH-1), the process jumps to step 2-6 and repeats steps 2-8.

In step 2-9, the variable X is (WIDTH-1
), 1 is added to the variable y (step 2-10). Here, y is (HE I
GET-1) is checked (step 2-11). If y is smaller than (HE I GHT-1), return to step 2-5 and step 2-10
Repeat the steps up to this point, and when y becomes equal to or larger than (HEIGHT-1), proceed to step 2-12. At this point, the processing for the entire read image ends.

On the other hand, in step 2-6, if the value of the referenced (called) binary image is "1", as shown in FIG. Check (step 2-13). That is, f (x-1,)') -1■ and f (x+1.V) -1■ and f (x, y-1) -1■ and f (x
, y+1)-1■. If the above formula is not satisfied, 1 of the pixel of interest
The next pixel is a normal pixel.

In other words, a pixel with a density value of 0 is adjacent to a normal pixel.

Therefore, in this case, following step 2-13,
Set the converted image g (x, y) to 0 (step 2-
14). Through this step, the first binary image memory 1
Even if the value is stored as ``1'' in 9, it will be converted to ``0'' if all formulas 1 to 2 are not satisfied. This indicates that the image will be deleted as a result of the narrowing process. This result is stored in the second binary image memory 22.

Next, the number of narrowing processes n is assigned as the distance image d (x, y) (step 2-15). This is step 2-
In 13, since the adjacent pixel is "0",
By narrowing processing, the pixel of interest (x.

y) disappears, the number of narrowing processes up to now is reduced from the pixel of interest to the normal pixel (threshold processing circuit 13).
This is the distance to a pixel that is clearly determined to be a background or printed part. Next, set the operation flag f to "1"
(Step 2-16). After this, step 2
Return to -8.

On the other hand, in step 2-13, if conditions (1) to (2) are satisfied, the converted image g (x, y) is set to "1" (step 2-17). This is because the pixels surrounding the pixel of interest are also "1", so the pixel of interest does not become a boundary of the binary image due to the previous thinning process. In this case, the transformed image g (x, y) is still “1”
I'll leave it as that. After this step 2-17, the process returns to step 2-8.

In step 2-12, the converted image g (x.

Copy the value of y) to f (x, y) (Step 2-
12). When subsequently converting to a distance image, the replaced f (x, y) becomes the original image. After that, 1 is added to the number of times n of narrowing processing (step 2-18). Furthermore, it is checked whether the operation flag f is 1 (step 2).
-19). If the operation flag f is 1, the process returns to step 2-3.

As mentioned in the explanation of step 2-16, the operation flag f is set to 1 when the thinning process is continued, and this means that depending on the current thinning process, the binary image This is because the "1" inside is not deleted.

In step 2-19, if the operation flag f is not "1", that is, if it is "0", this indicates that the entire image has been converted into a distance image. With this, the process of step 2 shown in FIG. 2 has been completed.

For example, in the case of an input image as shown in FIG. 7, the contents of the binary image memory 19 are as shown in FIG. By performing the above processing on this binary image, a distance image as shown in FIG. 9 is obtained. In the case of a binary image as shown in FIG. 8, blurring occurs on the right side of the number "0". Correspondingly, in FIG. 9, a portion of distance image "2" is generated.

After this, the process moves to counting specific pixels. As shown in FIG. 10, first, the number p of defective pixels is set to "0" as a variable (step 3-1). Next, the image vertical position y is set to "1" (step 3-2). Image horizontal position
is set to "1" (step 3-3).

Here, the distance MiifI image d(
x, y) are compared with a predetermined threshold MT)IR (step 3-4). The distance image d (x, y) is the threshold value TH
If it is larger than R, 1 is added to the variable p (step 3-5). By comparing with this threshold value, the one that is farther away from the image boundary is selected.

Next, "1" is added to the image vertical position X (step 3-6). On the other hand, if the distance image d (x, y) is equal to or larger than the threshold TIIL in step 3-4, the process directly proceeds to step 3-6. No. 9 above
In the case of the figure, the threshold value TI (L −1 is set).

Thereafter, it is checked whether the image vertical position X is smaller than (VIDT)I-1) (step 3-7).

If X is less than (WIDTH-1), step 3-
Return to step 4 and repeat the following steps. This completes the inspection of the horizontal portion of the target image.

In step 3-7, the image vertical position
If it is equal to or larger than I-1, then "1" is added to the image vertical position y (step 3-8).

Then, this image vertical position y is (HEIGHT -1)
Check whether it is smaller (step 3-9). If the image vertical position y is smaller than (HEIC)IT -1), return to step 3-3 and repeat the subsequent steps. In step 3-9, the image vertical position y is set to ()IP
If it is equal to or greater than IGHT~1), then the vertical inspection of the target image has been completed.

Next, the process moves to step 4. First, the variable p obtained above is compared with the determination value (step 4-1). When the variable p is larger than the judgment value, that is, the distance plane @d (x,
If the number of pixels with large y) is larger than the determination value, it is determined that there is a defective pixel (step 4-2). In other words, “
The judgment output is “defect”.

On the other hand, if the variable p is smaller than or equal to the determination value in step 4-1, a determination is made that there is no defective pixel, that is, "normal" (step 4-3). With the above processing, the image inspection has been completed.

Other Embodiments Next, a second embodiment will be described with reference to FIG. An input image from the image input device 11 is supplied to a mask processing circuit 101 . The signal output from the image input device 11 is a multivalued signal representing a grayscale image, as in the above embodiment.

The mask processing circuit 101 performs contour enhancement processing on the input image signal. As specific processing, Laplacian processing is performed as described later. This edge-enhanced image signal is converted into a binary image by the third threshold circuit 103 and sequentially stored in the binary image memory 105.

On the other hand, the multivalued signal output from the human-powered image device 11 is processed by the fourth threshold circuit 107 using two threshold values TL and TH, similar to the threshold circuit 13 of the first embodiment. Images that are not included in the background part and the printed part are extracted as binary images. This result is stored in the binary image memory 109
is memorized.

During the above processing, control such as timing is executed by the memory control circuit 111. This memory control circuit 1
11, a control clock is supplied to the binary image memories 105 and 107 as the image input device 11 operates.

On the other hand, the inspection device of this embodiment also includes CPUll3. This CPU 113 will be described later in accordance with a program stored in the program memory 115. Execute predetermined processing. When the storage of the binary signals into the binary image memories 105 and 109 is completed, processing associated with image input is shifted to processing by the CPU 113. During processing by the CPU 113, the binary image memories 105 and 107 are
It comes under the control of U113. Further, data and signals generated during processing by the CPU 13 are stored in the working memory 117. Further, the results of the specific processing performed by the CPU 113 are stored in the third binary image memory 118. The determination result as a result of the processing by the CPU 113 is supplied to the determination output circuit 119 and output to the outside.

Details of the mask processing circuit 101 will be explained according to FIG. 12. A multivalued image signal, such as an 8-bit serial signal, output from the image input device 11 is supplied to a j-line buffer 201.

This j-line buffer 201 stores signals for three pixels. For example, the 3 pixels on the j line (
Taking 3 pixel signals f (i, j), (i-1, j) and (i-2, j) as an example, f (i-1
, j) and f (i-2, j) are stored simultaneously.

On the other hand, all the signals constituting the (j-1) line have already been output from the image input device 11 when the j-th line data is input, and are stored in the (j-1) line buffer 203. ing. This line buffer 203
Writing and reading control to the j-line buffer 20
1, and is done as follows.

That is, now the pixel signal corresponding to f (i, j) is
When supplied from the input device 11, the pixel signal f (i, j-
1) is called and (j-1) line shift register 2
05 and the latch circuit 207. (j-1
) Data is written to the line shift register 205 while being sequentially shifted to the right. Latch circuit 2
The data written in 07 is stored in the (j-2) line buffer 209.

Above line shift registers 201, 205 and 211
, a specific bit group, that is, data in byte units is read out. That is, line shift register 2
2nd byte of 01, (j-1) line shift register 2
The first and third bytes of 05 and the second byte of the (j-2) line shift register 211 are read out and supplied to the adder 213, respectively. This adder 213 adds the four inputs, and outputs the result.

This output is expressed, for example, in 11 bits.

The output of this adder 213 is added with an offset, for example 128, in an adder 215. The output of this adder 215 is expressed in 12 bits. on the other hand,
(j-1) The second byte of data is read from the line shift register 205 via the 2-bit shift circuit 217.

By this 2-bit shift circuit 217, the data of the second byte of the (j-1) line shift register 205, that is, the data corresponding to f (i-1, j-1), is multiplied by four. This data is subtracted from the output of adder 215 in subtracter 219 . The output of this subtracter 219 is 12
The data is output in bits, and the limiter 221 outputs it as 8-bit data. This limiter 221 compares the input value with a predetermined value, in this case 255 due to the 8-bit limit, and outputs the smaller one. This output image is (
This is the Laplacian for the pixel i-1, j-1).

Now, to support the mathematical expression of the above processing, the adder 21
3 and 215 and the subtractor 219 are g (i-1, j-1) -128+ (f (i-1, j-2) + f (i-
2, j-1) 10f(i, j-1) +f (i-1, j) 1 -4f (i-1, j-1). Note that in the Laplacian processing in this embodiment, the first
As shown in Figure 3, for a 3×3 pixel group, the central pixel of interest (in the above example, (i-1°J-1)) is compared with the sum of the four surrounding pixels. Executed as a process.

Further, the processing by the limiter 221 is expressed as h (i-1, j-1) -min (255, g (i-1, j-1)). min (A, B) is a function that selects the smaller one of A and B.

Furthermore, three shift registers 201, 205 and 211
The storage state of data in the two line buffers 203 and 209 will be explained based on FIGS. 14 to 16. The control circuit 223 is in charge of this control, as will be described later. 14 and 15 represent pixels of data stored in shift registers 201, 205 and 211. FIG. 14 is a diagram created with the coordinates on the object to be read in mind, and the coordinate values become larger as you go to the right or down.

In the figure, the lowest row (i-2, 3), (i-1゜j),
Data corresponding to the pixel (i, j) is stored in the j-line shift register 205.

Similarly, the center row is (j-1) line shift register 2
05, the top row is (j-2) line shift register 21
1, respectively. shift register 201
．． FIG. 15 is a redrawn version of 205.211 along the shift direction.

The contents of both FIG. 14 and FIG. 15 are the same.

Next, among the above controls, the operation control of the (j-1) line buffer 207 will be mainly explained according to FIG. The control circuit 223 generates and outputs, as an operating clock, a data reference clock and an 8x data reference clock having a frequency eight times the frequency of the data reference clock. The control circuit 223 supplies and outputs address data for the N-1) line buffer 203 in synchronization with the rise of a predetermined data reference clock. When the pixel of interest is (i, j), this address data corresponds to (1, J-1).

For this address data, the control circuit 223 sends (j
-1) Supply a read signal to the line buffer 203. As a result, the data corresponding to (it j−1) is read from the line buffer 203, and the data corresponding to (j
-1) Line shift register 205 and latch circuit 20
8.

Note that the address in the (j-2) line buffer 209 corresponding to the pixel (t, j-2) is supplied to the (j-2) line buffer 209 at the same timing as above, and A memory read signal is also supplied to the (j-2) line buffer 209. Therefore, in response to the supply of the leftmost data reference clock shown in FIG.
j-1) and (j-2) line buffers 203 and 20
9, the pixel signals corresponding to (i, j-1) and (f, j-2) are respectively read out. Latch circuit 2
07, the above data read from the (j-1) line buffer 203 is latched. Next, in synchronization with the fifth 8x data reference clock with respect to the leftmost data reference clock, the data to be written next is written to (j-1) and (j-2) line buffers 203 and 209 and j ,
(j-1) (j-2) Line shift register 201.2
05 and 211. Specifically, (i,
The pixel signal corresponding to j) is sent to the j line shift register 2.
01 and (j-1) line shift registers 203. (j-2) For the line buffer 209,
A pixel signal corresponding to (i, j-1) is supplied via the latch circuit 207. The (j-1) line shift register 205 is supplied with a pixel signal corresponding to (i, j-1) from the (j-1) line buffer 203 .

(j-2) For the line shift register 211, (
j-2) From Reinbacher 209, (i.

A signal pixel corresponding to j-2) is supplied.

Next, a memory write signal is sent from the control circuit 223 to the line bachers 203 and 209, and the shift register 201.2.
05 and 211. This causes the shift registers 201, 205 and 211 to
Data is stored as shown.

Next, according to FIG. 17, the operation of the entire device is controlled by the CPU1.
The explanation will be centered on the processing in step 13. Image input device 11
When an image is input by , first, a contour is detected, and the contour image is binarized (step 21). The mask processing circuit 101 is used for the processing in step 21. In the same circuit 101, the outline of the input image is detected as described above. On the other hand, comparison with the threshold value THR is performed in the fourth threshold circuit 103. If the contour image is larger than the threshold THR, the fourth threshold circuit 1
03 outputs "1", for example. This binary signal (hereinafter referred to as a contour binary image) is stored in the binary image memory 105, and the collection is completed.

Next, the density value between the background area (paper portion) and the print area is extracted (step 22). This extraction is realized by separating the background area and the print area. This process
This is the same as the embodiment of , and is realized using two thresholds TL and TI. The results of this processing are sequentially stored in the binary image memory 109. Hereinafter, the signal binarized in this way will be referred to as an intermediate density binary image.

Once the 2N type binary image as described above is obtained, the contour binary image is deleted from the intermediate density binary image (step 23).
). This result is stored in the third binary image memory 11. Next, from the intermediate density binary pixels subjected to the deletion process in step 5-3, the remaining pixels, for example "1",
is counted (step 24). Based on this counting result,
A printing defect is determined (step 25). This judgment is
If the counted residual image is greater than or equal to a predetermined value, it is determined that the printing is defective, and if it is smaller than the predetermined value, it is determined to be good. Next, with reference to FIGS. 18 and 19, among the above processes,
The processing from step 23 onwards, that is, the processing of the CPU 113 will be explained.

First, 11 j is used as a variable, and j is set to "0" (step 22-1) and i is set to "0" (step 22-1).
-2). These variables are for coordinate settings.

Next, it is checked whether the intermediate density binary image is "1" for each b+(i,'j) of the contour binary image (step 22-3). This can be easily achieved by comparing the two binary image memories 105 and 109, but of course only pixels for which "1" is set as the contour image are called as h1 (11j). For this h+ (i, j), if the intermediate density binary image at the corresponding coordinates is "1", the process moves to outline deletion work.

Therefore, the variable k is set to "-2" (step 22-
4). Next, set variable 1 to “-2” (step 2
2-5). Next, among the contour images, those whose corresponding intermediate density binary images are "1" are converted into contour/defect binary images hz (
It is written as m+n). Then, here, h2 (i
+1. j+k) to "0" (step 22-6
).

Then, "1" is added to variable 1 (step 22-7
). Check whether the added l is equal to "3" (
Step 22-8). If 1 is not equal to "3", the process jumps to step 22-6 and the subsequent processing is repeated.

On the other hand, if variable 1 is equal to "3" in step 22-8, "1" is added to the variable k (step 22-9). Next, it is checked whether the variable k is equal to "3" (step 22-10). If the variable k does not reach "3" and is not equal, the process returns to step 22-5 and the subsequent processing is repeated. In step 22-10, if variable k is equal to "3", "1" is added to variable i (
Step 22-11). This added i is the image width V
Check whether it is equal to IDTH (step 2
2-12). If the variable i is not equal to the image width, the process returns to step 22-3 and the subsequent processing is repeated. If the variable i is equal to the image width, "1" is added to the variable j (step 22-13). Then, the added variable j is the image height H
EIG) and IT (step 22-14). If the variable j is not equal to the image height (IEIGHT), the process returns to step 22-2 and the subsequent processing is repeated. Through the above processing, the intermediate density binary image is deleted from the contour binary image.

Next, the process moves to counting the remaining contour binary images. As shown in FIG. 1, first, the number C of defective pixels is introduced as a variable and set to "1" (step 24-1). Next, each variable JSl is set to "0" (step 24-
2.24-3).

It is checked whether the contour/defect binary image h2 (i, j) is "1" (step 24-4). If h2 (i, j) is “1”, add []” to variable C (step 2
4-5). If h2 (i, j) is not equal to "1", skip step 24-5 and proceed to step 24-6.
to move to. In this step, "1" is added to the variable i.

The added l is compared with the image width vIDTH (step 24-7). If l is not equal to the image width V I DT)l, the process returns to step 24-4 and the subsequent processing is repeated. If i is equal to the image width V I DTH, "1" is added to the variable j (step 24-8). The added variable j is compared with the image height HEIG)IT (
Step 24-9). Variable j is image height IIEIGII
If it is not equal to T, the process returns to step 24-3 and the subsequent processes are repeated. If the variable j is equal to the image height] (EIGHT, the counting of remaining pixels is completed, and the process moves on to determining the quality of printing.

First, it is checked whether the variable C obtained above is equal to a reference value (step 25-1).

If variable C is not equal to the reference value, it is determined that printing is normal,
A message to that effect is output (step 25-2).

If the variable C is equal to the reference value, it is determined that defective printing exists, and a message to that effect is output (step 25-3). With this, the inspection process ends.

Although the embodiments of the present invention have been described above, the present invention is not limited to the same extent as in the above embodiments. For example, the reading means may be a color one, and the device used is not limited to a CCD or the like. In addition, boundary detection and threshold selection are also possible.
Executed on a whim.

Furthermore, a wide variety of methods are used to separate defective pixels from boundaries, such as a distance measurement method and sequential deletion of boundaries.

[Effects of the Invention] As described above, according to the present invention, it is possible to successfully inspect images such as printed images without preparing a standard pattern.

[Brief explanation of the drawing]

FIG. 1 is a diagram showing the configuration of an image inspection device, FIG. 2 is a diagram for explaining the determination of the threshold value shown in FIG. 1, and FIG. 3 is a diagram showing the main processing of the device shown in FIG. 1. 4 is a diagram for explaining the conversion of the step intermediate distance image shown in FIG. 3, and FIGS. 5 to 10 are diagrams for explaining the details of the process shown in FIG. 3. FIG. 11 is a diagram showing the configuration of the apparatus according to the second embodiment. FIG. 12 is a diagram showing details of the mask processing circuit shown in FIG.
3 through 15 are diagrams for explaining the operation of the circuit shown in FIG. 12, FIG. 16 is a diagram showing signal waveforms in the circuit shown in FIG. 12, and FIGS. 17 through 19. The figure is a diagram showing processing in the apparatus shown in FIG. 11.

Claims (2)

[Claims] (1) For the object to be inspected on which an image is provided,
An image inspection method for inspecting the condition of the image includes: a reading means for reading the surface of the object to be inspected including the image and converting it into a corresponding signal; means for separating and extracting a signal containing a signal; means for determining the distance from a pixel corresponding to the signal separated by this means to the outline of the image; and inspecting the image based on the distance obtained by this means. An image inspection method characterized by performing the following. (2) For the object to be inspected on which an image is provided,
The image inspection method for inspecting the state of the image includes: a reading means for reading the surface of the object to be inspected including the image and converting it into a corresponding signal; and a lower contour corresponding to the contour of the image from the signal converted by the means. a detection means for detecting a signal; a means for separating and extracting a signal including a signal caused by a defect in the image from the signal read by the reading means; 1. An image inspection method comprising means for deleting a contour signal, and inspecting said image based on the remaining signal after deletion by said means.