JPH05120435A - Luminance signal comparing and judging method - Google Patents

Abstract

PURPOSE:To reduce an error by weighting a boundary line, at the time of comparing and judging a video signal. CONSTITUTION:The time of a signal IN to be judged is delayed in a V-direction, a coincidence or non-coincidence between the center signal of a delay time and three signals before and after the delay is judged, and the time of the delay signal from the middle point is delayed in an H-direction. The coincidence and anti-coincidence between the middle point signal of the delay time in the pertinent direction and the three signals before and after the delay is judged, and the judgement signal of the coincidence and non-coincidence of the three signals and the previous judgement signal of the coincidence and non-coincidence of the three signals are used as weighting signals. Then, a reference signal REF is delayed in the V-direction in the same delay time as the middle point of the above mentioned delay time in the V-direction, and it is delayed in the H-direction in the same delay time as the middle point of the above mentioned delay time in the H-direction. The pertinent delay reference signal is compared with the signal from the middle point of the second delay time for the above mentioned discrimination signal, and the output is weighed by the above mentioned weighting signal, and the similarity is judged by a comparator 3.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention generally relates to analogy discrimination of luminance signals, that is, video signals. In particular, the accuracy of the discrimination is improved by reducing various influences that cause the discrimination error of the luminance signal. That is, in a method of determining whether or not the discriminated signal is similar to the comparison signal by comparing the discriminated signal which is a video signal discriminated through a digital circuit with a comparison signal to be compared. The present invention relates to a method for increasing the accuracy of discrimination by weighting the degree of discrimination near the threshold value for binarizing an analog video signal or excluding the weighted portion.

[0002]

2. Description of the Related Art There are various methods for determining whether or not a certain object has the same appearance. Among them, a method is often used in which the object is shown as an image signal in an image, the two images are discriminated to be coincident and discriminated, and the analogy of both images is discriminated. In this case, the method of multiplying the video signals of both images by an electronic calculator has been used in the past, but this not only results in a very high cost, but also because it takes time to determine the analogy and improves accuracy. It became even more expensive.

However, there are many demands for discriminating whether the shapes or patterns of objects are the same or not. For example, there are various kinds such as whether or not the parts of the product are properly attached, whether or not the characters or the like printed on the IC are correctly printed, and whether or not the stamp imprinted on the document is genuine. In particular,
In the case of a fine shape, it is easy for the human eye to make a mistake and cannot continue physically or mentally for a long time. In particular, temporal follow-up is also a problem in distinguishing products that are sent one after another on a belt conveyor or the like. Therefore, an invention was first made in which a certain object was first processed as a video signal and solved by using the above-mentioned image pickup method. That is, first of all, if a reference object is regarded as a video signal and compared with a video signal from an object to be discriminated, whether the video signal from the object to be discriminated matches the above-mentioned video reference signal. It is possible to determine whether or not. That is, by comparing these two video signals, the degree of coincidence of both video signals, that is, the degree of coincidence of the appearances of both objects can be known, and the similarity can be determined. Then, in this case, the video signals of both are normally compared with the reference value through the integrating circuit for the absolute value of the difference.

[0004]

However, in the above, first, the accuracy of the similarity judgment is hindered by the noise due to various causes of the external device, the image pickup device itself and the like. Secondly, in the case of comparing both video signals as described above, a problem occurs in the case of a binarized signal. That is, in the case of a binarized signal, an analog signal is obtained by A / D conversion, but the analog signal is divided into signals of 0 or 1 with a constant threshold value as a boundary. Therefore, an analog signal that is close to the threshold value is evaluated as it is in the case of analog, but is 0 in digital.
Misunderstanding occurs because it is forced to divide into one extreme. Specifically, this occurs in a portion where there is little difference in luminance near the threshold value, a boundary between image shapes and patterns, and the like.

[0005]

SUMMARY OF THE INVENTION In view of the above points, the present invention firstly reduces the adverse effects of external noise and internal noise caused by the image pickup device itself, etc., in order to reduce the adverse effects of the adverse effects. Weighted accordingly. Note that this weighting includes 0, and is the same in the following description. Secondly, at the luminance signal exchanging stage, the comparison result is given a stage value, that is, a weight around the reference point for exchanging, so as to avoid the adverse effect of a portion which causes a misjudgment. This error that occurs when A / D conversion of this analog signal is given a certain width near the threshold as a method that does not affect the judgment as much as possible, and the comparison result of the binarized signal within that width The above problem is solved by performing weighting including 0 in. Thirdly, a delay time is provided in both the V and H directions of the binarized signal of the image, and the evaluation value of the similarity determination is reduced at a portion where the signals in the binarized signal do not match.

The specific configuration of the luminance signal comparison / discrimination method according to the present invention will be described below. First, by comparing a discriminated signal which is a video signal discriminated through a digital circuit with a comparison signal to be compared, a method for discriminating whether or not the discriminated signal is similar to the comparison signal is described. Assumption. Therefore, the absolute value of the difference between the signal to be discriminated and the comparison signal (reference signal) is output. The portion with a low absolute value is weighted, that is, it is excluded without being evaluated as it is or its value is reduced. Thereafter, this weighted signal is integrated and compared with a signal serving as a reference for comparison of similarity in a comparator, which is a luminance signal comparison / discrimination method.

Then, first, the analog discriminated signal obtained from the imaging method is converted into a binarized discriminated luminance signal. Then, the comparison signal obtained from the storage method or the like is also converted into a reference luminance signal for binarization. On the other hand, a certain width is provided above and below the binarization threshold of the signal to be discriminated, and the absolute value of the difference between the two signals within that width is weighted, integrated and then compared. The signal is compared with the signal that serves as a reference for similarity. This is a luminance signal comparison / discrimination method configured as described above.

Finally, the conviction of the present invention will be described. First, the signal to be discriminated is delayed in the V direction. A signal centered on the center of the delay time, three signals including a signal before the delay and a signal after the delay are taken, and the coincidence or non-coincidence of the three signals is determined. Then, the delay signal from this midpoint is delayed in the H direction. A signal centered on the midpoint of the delay time in this direction, and three signals of the signal before the delay and the signal after the delay are taken, and the coincidence / non-coincidence of the three signals is discriminated. Then, the determination signal of the match / mismatch of the three signals and the determination signal of the match / mismatch of the three signals are used as the weighting signals.

On the other hand, the reference signal is taken from a storage device or the like. This reference signal is delayed in the V direction by the same delay time as the midpoint of the delay time in the V direction. Next, this is delayed in the H direction by the same delay time as the midpoint of the delay time in the H direction.
The delayed reference signal is compared with the signal from the midpoint of the second delay time for the discrimination signal. Then, this output is weighted by the weighting signal.
Then, the weighted signal is compared with a signal serving as a reference for similarity by a comparator to determine the similarity. This is a luminance signal comparison / discrimination method configured as described above. The configurations of the signal to be discriminated and the reference signal may be positioned opposite to each other. Further, the above-mentioned delays of V and H may be positioned in reverse.

[0010]

The function according to the present invention has the following effects because of the above-described configuration. First, the absolute values of the signal to be discriminated and the reference signal are output. Since the portion with a low absolute value has a lot of noise components, the signal obtained from this portion, that is, the error signal is weighted. Accordingly, a luminance signal comparison / discrimination method in which the coincidence / non-coincidence determination error is reduced is performed. Next, an analog video signal to be discriminated is obtained by the imaging method. This is binarized. Then,
A certain width is provided near the binarization threshold value, and the signal within the width is not used as a target for comparison / discrimination as described above, or its value is evaluated low, so that the determination error there is reduced. did.

Finally, in the present invention, first the signal to be discriminated is V
By delaying the time in both the directions H and H, three signals having the widths of the signal, the signal before the delay and the signal after the delay are obtained with the center of the delay time as the center. On the other hand, there is a reference signal, which is delayed by the same delay time as the midpoint of the delay time. Then, the output from the delay circuit for the reference signal and the output signal from the midpoint of the delay circuit for the discrimination signal are input to the comparator. When the above three signals match, the output from this comparator is output as it is, and when they do not match, the output signal is excluded or reduced and evaluated. Specifically, the output from the comparator and the coincidence / non-coincidence signal of the three signals are input to the weighting circuit. Therefore, instead of this weighting circuit, a switch circuit that operates with the above three signal coincidence and non-coincidence, and switches the output from the above comparator to a value with a high or low stage including 0. It may be. Based on the above, this luminance signal comparison / determination method reduces the determination error. The circuit at the final stage is not limited to the circuit described above as long as it is a signal having a value including high and low stages including 0.

[0012]

BEST MODE FOR CARRYING OUT THE INVENTION The above-described luminance signal comparison / discrimination method will be described in detail below with reference to the accompanying drawings showing each of the embodiments. First, by comparing a discriminated signal which is a video signal discriminated through a digital circuit with a comparison signal to be compared, a method for discriminating whether or not the discriminated signal is similar to the comparison signal is described. Assumption.

Thus, as shown in the schematic circuit diagram of FIG. 1, the signal to be discriminated is input to IN and the reference signal is input to input REF. Then, the absolute value of the difference is taken out through the schematic circuit 1 showing an example. Next, this is circuit 2
Cut or attenuate the low part of the signal, etc. Of course, this attenuation or cut may use the characteristic of the diode used when taking out the above-mentioned absolute value, that is, the characteristic that the conductive characteristic is changed at about 0.6 V in the case of a silicon diode to have a rectifying action. Then, the weighted signal is integrated and compared with a signal serving as a reference for similarity by a comparator to determine the similarity. This is the method for comparing and determining the luminance signal based on the above.

Next, there is an imaging method for taking an image from the object to be discriminated. A to-be-determined signal taken by this imaging method is
It is sent to the / D converter, where it is converted into a binary signal and sent to the comparator. On the other hand, the comparison signal is obtained by the comparison image means. The comparison image means may be one in which a reference object is directly captured by an imaging method, or one in which various recording methods such as those recorded in a video recorder are reproduced and used. Next, the signal from the comparison image means is sent to another A / D converter, where it is converted into a binarized signal. Then, the output is sent to the comparator. The two signals entering the comparator are compared here and the output signal is sent to the admission means. The video signals from the discriminating imaging method and the reference imaging method cannot be 100% identical under any condition due to internal and external factors. Therefore, a certain degree of tolerance is required to determine whether the two signals match or mismatch. This allowance is set by this allowance means, and only those that are within that allowance range are judged to be coincident.

However, in the case of digital processing, "0" or "1" is still generated near the binarization threshold value as described above.
Will be divided into Therefore, the error in the vicinity increases. Therefore, in the present invention, a constant width is provided above and below the thresholds of both A / D converters, and when there is a luminance signal within that width, the error signal generated within that width is weighted. .. Then, the weighted signal is integrated and compared with a signal serving as a reference for similarity judgment by a comparator.

A part of the above circuit is shown in FIG. 2 in a concrete circuit diagram. In the case of the embodiment shown in FIG. 2, the weight is 0 because it is a digital signal. First, the binarized discriminated signal for discrimination obtained as described above is sent to the digital signal comparator 3 from the input IN. On the other hand, the reference signal, which is a comparison signal, is binarized, inputted from the input REF, and sent to the digital signal comparator 3, where the two signals are compared. Here, 20 is a first threshold setting device, 21 is a second threshold setting device, and 22 is a third threshold setting device. And 30, 31, 32 are A /
It is a D converter.

The binarized signal to be discriminated, which is input from the input IN, is separated from the binarized signal and passes through the comparators above and below the threshold value to set the weighting width. It is obtained from the circuit 7 and enters the weighting circuit 8. The output from the digital signal comparator 3 is also input to the weighting circuit 8. Therefore, the error signal having the width determined by the weighted width setting circuit 7 is not output from here. Therefore, the error in that range can be reduced. If the switch circuit 8a shown in FIG. 3 is used instead of the weighting circuit 8, the switch is switched by the weighting width setting circuit 7 and the output signal from the digital signal comparator 3 is output. It becomes possible to give the degree of coincidence degree a light weight.

However, as is common to all the inventions in the present invention, it is possible to determine the lightness / multiplicity value of a signal in a plurality of stages by using a plurality of stages of those circuits. Finally, the third invention will be described. First, the signal to be discriminated enters from the input IN. This discriminated signal is delayed by H and V
Delay the time in both directions. This delay circuit is specifically a series of one pixel delay circuit 9 in the V direction coupled in series.
And 10 and the midpoint 11 thereof, and one pixel delay circuit 12 and 13 in the H direction connected in series. Then, a signal centered on the midpoint of the delay time and
Taking three signals, the signal before the delay and the signal after the delay,
The match / mismatch of the three signals is discriminated by an AND, NAND, OR circuit or the like. Specifically, the horizontal weighting circuit 14 and the vertical weighting circuit 15 are used. Then, both outputs are input to the OR circuit, and the output is used as a weighting signal.

On the other hand, the reference signal comes in from the input REF. This reference signal is delayed in each of the V and H directions by the reference signal delay circuit 16 or the like having the same delay time as the midpoint of the delay time. The delay circuit 16 is specifically composed of a one-pixel delay circuit 16a in the V direction and a one-pixel delay circuit 16b in the H direction. Then, the output from the delay circuit 16 for the reference signal and the H for the determination signal
The output signal from the middle point of the directional delay circuit is compared with the comparator 1
Type in 7. Finally, the output from the comparator 17 and the output from the discriminant step value signal circuit are weighted by the weighting circuit 1.
Enter in 8. Then, the signal from the comparator 17 is output or stopped as it is according to the weighting signal.

Instead of the weighting circuit 18, however, as shown in FIG. 3, a switch circuit 18a which operates by the output from the above-mentioned discrimination stage signal, and which outputs the output from the above-mentioned comparator. The value may be switched to a value having a high or low level including 0. As described above, the weight may be changed by using a plurality of stages. This is a luminance signal comparison / discrimination method configured as described above. In addition,
The circuit at the final stage is not limited to the circuit described above as long as it is a signal having a value including high and low stages including 0. Then, the weighted signal is integrated and compared by a comparator with a signal serving as a reference for analogy judgment, and the analogy is judged.

By the way, as is common to all the inventions, the symbol to be discriminated may be obtained from an ordinary imaging method for visible rays, or may be obtained from an imaging method utilizing invisible rays such as X-rays. May be Further, the reference signal may be obtained directly by the image pickup method or may be obtained by a storage method such as a video tape recorder or a semiconductor memory. Further, the integration may be performed only for the window, that is, the required portion of the image.

[0022]

Due to the above-mentioned structure, the method for comparing and discriminating the luminance signal according to the present invention is excellent in that the discriminating error due to the boundary of the image, noise or the like is extremely reduced.

[Brief description of drawings]

FIG. 1 is a partial schematic circuit diagram used in an embodiment of a luminance signal comparison / determination method of the present invention.

FIG. 2 is a circuit diagram for use in the second invention.

FIG. 3 is a circuit diagram of an example of the switch circuit.

FIG. 4 is a circuit diagram showing a part of the embodiment.

[Explanation of symbols]

 1, 4 ... Imaging method 2, 5 ... A / D converter 3 ... Digital signal comparator 7 ... Weighted width setting circuit 8a ... Switch circuit 9, 10, 16a ... One pixel delay circuit 11 ... Midpoint 12, 13, 16a ... 1 pixel delay circuit 17 ... Digital signal comparator 20 ... 1st threshold value setting device 21 ... 2nd threshold value setting device 22 ... 3rd threshold value setting device 30, 31, 32 ... A / D converter

Claims (1)

[Claims] 1. A discriminated signal, which is a video signal discriminated through a digital circuit, is compared with a reference signal to be compared to discriminate whether or not the discriminated signal is similar to the reference signal. A) H of the signal to be discriminated or the reference signal
Alternatively, the signal is delayed in either direction of V or V, and a signal centered at the midpoint of this delay time, a signal before the delay, and a signal after the delay are taken, and the three signals are determined to match or not match. Simultaneously with the determination; delay the signal from the above middle point in the direction of H or V, which is not delayed in the above direction, and set the signal centered at the middle point of the delay time in this direction and 3 signals of a signal before delay and a signal after delay of a direction are taken, and a match / mismatch of 3 signals in this direction is determined; a match / mismatch determination signal of 3 signals in this direction and a match of 3 signals in the first direction The discriminant signal of non-coincidence is used as a weighting signal, while; b) the reference signal or the signal to be discriminated, which is not delayed above, is delayed by the same point as the midpoint of the delay time of the second delay described above. Time delay; this delayed signal and the above Comparison with the signal from the midpoint of the second delay time for the discrimination signal or the reference signal; this output is weighted with the above weighting signal and then integrated, and then the output is compared with analogy. Brightness signal comparison / discrimination method to compare with a detector.