JPS6343037B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to an image reading device that reads or recognizes pattern images such as characters and figures by binarizing them at a predetermined level, and in particular, an image reading device that reads or recognizes pattern images such as characters and figures at a predetermined level. The present invention relates to an image reading device configured to make a decision. Generally, in an image reading device that reads the target pattern by binarizing the video signal obtained by capturing the target pattern with an imaging device such as a television camera, It is desirable to be able to set an appropriate binarization level in consideration of variations on the surface of the image, or variations in characteristics of the imaging device due to ambient temperature.

However, conventionally, the binarization level is set to a predetermined fixed value, and the video signal obtained through the imaging device is binarized by comparing it with the predetermined binarization level. It is clear that variations in the video signal level cannot be accommodated due to changes in the video signal level. For this reason, measures must be taken to keep fluctuations in the lighting equipment that illuminates the object to be measured, such as fluctuations in its power source, to keep the color and brightness of the surface of the object to be measured constant, and to keep the ambient temperature constant. Efforts are being made to minimize fluctuations in the video signal level. In other words, a fixed 2 value that binarizes the target pattern image at a fixed level.
The value converting method requires the above-mentioned special measures, and even with such measures, there is a drawback that it cannot deal with fluctuations in the video level at all.

The present invention has been made in view of the above, and it is an object of the present invention to provide an image reading device that can determine an appropriate binarization level even when various fluctuations as described above occur.

The feature of this invention is that the operation of binarizing the target pattern by setting a different binarization level each time for the video signal obtained by imaging the pattern to be measured with a television camera or the like is performed a predetermined number of times. The point is that it is possible to cope with level fluctuations in the video signal by finding the case where the target pattern best matches a predetermined standard pattern and reading the target pattern using the optimal binarization level in that case. .

Embodiments of the present invention will be described below with reference to the drawings.

FIG. 1 is an overall configuration diagram showing an embodiment of the present invention;
Figure 2 is a block diagram showing the binarization device in Figure 1, Figures 2A to 2C are waveform diagrams of various parts to explain the operation in Figure 2, and Figure 3 is to explain the operation in Figure 1. This is a flowchart.

In Figure 1, 1 is a television camera, 2
3 is a binarization device, 3 is an image capture device, 4 is an image memory, 5 is an arithmetic control device, and 6 is a judgment result output device, and these are used to check the engine model stamped on the engine block for characters related to the model. Chietsuka is composed. Therefore, the object of inspection in this embodiment is the model stamp of the engine block, and the stamp is illuminated with specular reflection and photographed by the television camera 1. Here, regular reflection illumination is a method used when inspecting depressions on a flat surface, in which the inspection surface is observed from an oblique direction using an imaging device such as a television camera.
A diffused uniform illuminator (surface light source) is placed at a position that is a mirror image of the imaging device with the inspection surface as the center. In this case, if the inspection surface is close to a mirror surface, the imaging device will see the illuminator on the side opposite the marking.
This method utilizes the principle that since a mirror surface cannot be formed in areas with depressions, only those areas do not reflect light and appear black.

The video signal photographed by the TV camera 1 in this manner is given to the binarization device 2. Part 2
The digitization device 2 selects a binarization level based on the fixed binarization level selection signal from the arithmetic and control device 5, and converts the binarization level into two.
The digitized image signal is sent to the image capture device 3. The image capturing device 3 writes image data (segment length, right point coordinates, connection information, etc.) into the image memory 4 for each screen based on instructions from the arithmetic and control device 5. The arithmetic control section 5 performs predetermined arithmetic operations based on this image data to determine characters, and sends the results to the output device 6.

Here, the binarization apparatus shown in FIG. 1 will be explained with reference to FIGS. 2 and 2A to 2C.

In FIG. 2, 7 is an amplifier, 8a, 8b are delay elements, 9 is a fixed voltage source, 10 is a voltage divider, 11a
11c is a comparator, 12a to 12c are monostable multivibrators (hereinafter abbreviated as monostable),
13a to 13d are OR elements, and 14a and 14b are adjusting voltage dividers.

A video signal from the TV camera 1 is amplified by an amplifier 7, and the amplified video signal A is delayed by, for example, 1 μsec by delay elements 8a and 8b, respectively, to become video signals B and C. On the other hand, the voltage of the fixed voltage source 9 is divided into N ways by a voltage divider 10. Therefore, the voltage divider 10 outputs a predetermined voltage divided into a predetermined value in response to a binarization level selection command given from the arithmetic control section, thereby setting the binarization levels TV1 to TV9 (the (See Figure 2A)
Set. The video signal B and the binarization level output from the voltage divider 10 are compared by a comparator 11a, thereby performing binarization. moreover,
One of the signals is direct and the other is the monoste 12
Since the signal is applied to the OR element 13a via the OR element 13a, a binary signal with a predetermined pulse width is obtained from the output of the OR element 13a. For example, if the waveform of the video signal B is as shown in FIG. 2A, and the binarization level is set to TV5 for the video signal B, an output D as shown in FIG. It will be done. Note that _the binarized _output shown by the dotted line in FIG. The reason for doing this is to ensure writing to the memory by setting the binarized output pulse width to a predetermined size. On the other hand, the video signal A obtained through the amplifier 7 is divided by the adjusting voltage divider 14a and compared with the delayed video signal B in the comparator 11b.
A binary signal E is obtained by performing an OR operation on the output of 1b and the output obtained by making the output into a predetermined width by the monoste 12b using an OR element 13b. In other words, as is clear from Figure 2B, by taking the difference between the delayed video signal B and the video signal A divided to an appropriate value, the rising edge of the video level on the right side of the TV camera is emphasized. do. This is to emphasize the right border of the stamped characters. This method is not affected much by the constant rise and fall of the video level due to temperature and light intensity, so stable output can be obtained by appropriately adjusting the partial pressure ratio. However, the problem here is in the case of cut surfaces such as stamped surfaces, in which case the roughness of the surface may vary and cut marks may appear. Therefore, in order to read the markings when the cut surfaces are aligned in this way, it is desirable to adjust the partial pressure value so that the cutting marks do not appear as a binary signal.
The same holds true when emphasizing the left border versus emphasizing the right border. That is, as shown in FIG. 2C, a comparator 11c compares the adjusted output obtained by adjusting the voltage of the delayed video signal C with the voltage divider 14b and the video signal B (see A in the same figure). A digitized signal is obtained, and the signal is further converted into a pulse signal of a predetermined width by a monostage 12c to obtain a binarized signal F (see b of the same figure). The binary signals D, E, and F thus obtained are further logically summed by an OR element 13d and supplied to the image capture device 3.

Next, the operation shown in FIG. 1 will be explained with reference to FIG. 3.

First, when the power is turned on, the arithmetic and control unit 5 enters an initial state and starts processing (◯A). Normally, the "judgment start" signal is monitored (○B) and a predetermined routine work is executed (○D). Routine work includes monitoring displays or equipment abnormalities. Next, when a "judgment start" signal is received, the final binarization level in the previous determination operation is set (◯C), and an image capture command is given to the capture device 3 (◯H).
When the image data is written into the image memory 4, character reading is executed (◯). Based on the target pattern read in this way, the arithmetic and control unit 5 compares the target pattern with the standard pattern, and determines whether the distance between the target pattern and the standard pattern is within a predetermined range.
The degree of correlation or similarity of the target sample pattern to the standard pattern is determined (○). If the judgment result is within the set range, the judgment result is output (○), and if it is outside the set range, a binarization level selection command is sent to the binarization device 2 to a predetermined appropriate binarization sequence. Therefore, set it again (○chi). Here, the appropriate binarization series is a level series indicating a series of binarization level selection methods used when searching for a binarization level that allows accurate character reading by trial and error method. be.

FIG. 4A is an explanatory diagram for explaining an example of such a binarization level system. That is, in FIG. 4A, the first binarization level is set to TV5, and the second
The binarization level is sequentially changed, such as TV6 for the first time and TV5 for the third time, to find a binarization level that allows proper character reading. When it is determined that the reading has been performed properly, the binarization level at that time is stored as the final binarization level, and is used as the next binarization level.

FIGS. 4B and 4C are explanatory diagrams for explaining other examples of the binarization level series.

In the case of Figure 4B, it is suitable for use when the video signal fluctuates; the final (optimal) binarization level at the time of the previous judgment was TV1, and starting from this, the binarization level is sequentially set one by one starting from TV1. When the level reaches TV9, the user repeats the process of decreasing the level one by one, searching through trial and error for the optimum character reading.

In addition, Fig. 4C takes advantage of the fact that the minimum distance between the sample pattern and the set standard character feature amount is minimal between the binarization levels TV1 to TV9, and the character reading method is pattern matching. It may depend on the law. In this case, the distance between the sample pattern and the standard character is expressed by the sum of these mutually different pixels or line elements. Therefore, the distance is calculated by calculating the sum of the different pixels between the standard pattern of characters and the sample pattern,
The smallest one among them is defined as the minimum character distance. In other words, since the characters used as the engine model are limited, memorize the standard pattern of that character in advance, perform munching between the standard pattern and the sample pattern, and set the sum of different pixels for each character. The minimum character distance among the distances for each set character is determined as the minimum character distance, and the setting level at which the minimum character distance is obtained is determined to be the optimal level. For example, as shown in Figure 4C A, the first fixed binarization level is TV
3 is selected. This level TV3 may be, for example, the previous optimum binarization level, or may be a predetermined value that appears most frequently among the optimum binarization levels in the past several dozen binarization operations.
Level TV5, initially +2 from level TV3,
Next, the minimum character distance at that time is determined by setting the level TV1 which is decreased by -2, and if there is no minimum value in that section, the level TV7 which is increased by +2 in the diagonal direction is selected. If there is no local minimum in that section, the level TV9 which is further increased by +2 is determined and the local minimum range TV6,
TV8 is determined, and TV7, which is minimum, is determined last, and the determination is completed after confirming that it is minimum. Figure B shows the relationship between the binarization level and the minimum character distance in this case, and the numbers on the curve in the figure indicate the number of times the level is set.
The first time was at level TV3, the second time was at level TV5.
, and in the same way, set the level for the 8th time.
This is an example in which the minimum value of the minimum character distance was obtained by setting TV7. While the method shown in FIGS. 4A and 4B is a trial-and-error method, this method can also be called an asymptotic method.

As described above, according to the present invention, various disturbances (variations in the amount of light source, temperature, brightness of the surface to be inspected, etc.)
Appropriate 2 regardless of video level fluctuations caused by
Since the digitization level can be selected, a clear binarized image can be obtained as a result.

The present invention can be applied not only to engine block type (inscription) reading, but also to printed character reading devices and recognition devices in general that recognize patterns of fixed shapes.

[Brief explanation of the drawing]

FIG. 1 is an overall configuration diagram showing an embodiment of the present invention;
FIG. 2 is a block diagram showing a specific example of the binarization device in FIG. 1, FIGS. 2A to 2C are explanatory diagrams for explaining the operation of FIG. 2, and FIG. Particularly, the flow chart for explaining the operation of the arithmetic and control device, and FIGS. 4A to 4C are explanatory diagrams for explaining an example of setting the binarization level. Description of symbols, 1... Television camera, 2... Binarization device, 3... Image capture device, 4... Image memory, 5
... Arithmetic control device, 6... Judgment output device, 7... Amplifier, 8a, 8b... Delay element, 9... Fixed voltage source, 1
0, 14a, 14b...Voltage divider, 11a-11c...
Comparison circuit, 12a to 12c...monostable multivibrator (monoste), 13a to 13d...OR element,
TV1 to TV9...Binarization level.

Claims (1)

[Claims] 1. In an image reading device that reads a target pattern image by binarizing an imaging signal obtained by imaging a measurement target pattern using an imaging means at a predetermined level, a plurality of mutually different binarization levels are set. a level setting means, a binarization means for binarizing the target pattern based on the set level, and a target pattern by comparing the binarized target pattern by the binarization means with a predetermined standard pattern. determination means for determining the degree of correlation with respect to the standard pattern, and the setting means sets the binarization level each time to convert the target pattern into two.
An image reading device characterized in that a binarization level at which the degree of correlation becomes a predetermined value is determined by performing a digitization operation a predetermined number of times.