JPS61105682A - Graphic recognition processing device - Google Patents

Abstract

PURPOSE:To identify the pattern by clamping the first and second level signals outputted from a photoelectric conversion means and calculating the similarity between the standard pattern and the mean level signal of plural first level signals which tends to make level fluctuation due to the density difference of printing. CONSTITUTION:Comparative output (C<0>, C<1>) is inputted to a comparator 24 from a processing circuit 28. Comparative signals from an image memory 22 through a comparing means 11 are calculated of its similarity with the standard pattern of memory pattern of memory medium 1 in the nearer order (Sa Sb Sc) of mean level signal against the first level signal S1, and as the result, judgement is done whether the recognition of the recorded pattern is possible or not. If, as a result of calculation and judgement, the recorded pattern is judged to be recognizable, the recognition processing is executed to define what the recorded pattern is.

Description

[Detailed Description of the Invention] (Field of the Invention) The present invention is useful for determining whether a betting ticket such as a horse racing ticket or a car ticket is a winning ticket, or for reading an entrance ticket, a cash ticket, etc. Alternatively, the present invention relates to a figure recognition processing device used to read figures such as English characters.

(Summary of the Invention) The present invention provides a method for reading and recognizing numbers, etc. on a voting ticket, for example, by a photoelectric conversion means through a lens system, regardless of the arrangement pattern of the figures. In order to obtain a faithful signal and to quickly perform figure recognition processing regardless of the difference in shading of the printed figure,
Predetermined processing is applied to the output of the photoelectric conversion means and the figure recognition processing.

(Prior art and its problems) As shown in Figure 2, this type of device uses a recording medium as an object to be read/read! For a figure A such as the number above, the first reflected light from the recorded part of the figure A and the second reflected light from a place other than the recorded part are converged via the lens system 2, and the COD It is configured to form an image on a photoelectric conversion means 3 such as a linear image sensor, and output it as a first level signal and a second level signal, respectively, depending on the level of the reflected light. The first and second level signals are then input to the amplification/A/D converter 4, where the analog signals are binarized or multi-valued and converted into digital signals. It is configured to recognize the figure A by matching it with a figure pattern.

The conversion of the above-mentioned analog signal into a digital signal is performed, for example, at the bright level (generally, the level of reflected light from the paper as a recording medium!) and the dark level (generally, the printed figure A).
Although A/D conversion is not necessary if each of the reflected light levels (reflected light levels) is approximately constant, it is actually necessary due to the following factors.

(i)'PCS value (contrast value) becomes about 0.5 for a lightly printed figure.

(11) Due to poor uniformity of the light source, the amount of light at both ends of the recording medium l tends to decrease.

(iii) The lens system 2 tends to reduce the level of reflected light from the outer periphery of the recording medium 1.

(iv) The output of the photoelectric conversion section is not uniform.

(V) Changes in the reflectance of the mount such as the paper quality of the recording medium I.

Therefore, in the past, analog signals were converted into digital signals using a configuration as shown in FIG.

That is, first, since the first and second level signals output from the photoelectric conversion means 3 are separated pixel by pixel, it is not possible to detect and output the second level signal as it is. The level signal is amplified by an amplifier 06 and then sampled and held by a sample hold circuit 07.

The output (a) from the sample hold circuit 07 is input as is to the voltage comparator 08, and the output (a) is connected to a first peak value detection circuit 09 for outputting the bright level envelope and a dark level envelope. The input signal is input to each of the second peak value detection circuits 010 for outputting an envelope, and the envelope is output from each of the first and second peak value detection circuits 09 and 010 (b), (
c) is impedance-converted through an impedance converter 011 and input to the setting means 012, and an intermediate level signal (d) intermediate between the first level signal and the second level signal is set and output. There is.

Furthermore, the intermediate level signal (d) is input to the voltage comparator 08, and the output from the sample and hold circuit 07 (a) is compared with the intermediate level signal (d) to correspond to the first and second level signals respectively. A binarized comparison signal is output.

However, when scanning is carried out along the line L1 as shown in the upper part of FIG. The waveform will be as shown in (a).

On the other hand, in the case of the conventional example with the above-mentioned configuration, when the time constant (Ct・Rt) due to the capacitor and resistor of the second peak value detection circuit 01G is made small, the envelope of the second level signal The line output (c) is as shown in Figure 8,
There is a tendency to increase between adjacent peak values, and accordingly, the intermediate level signal (d) changes even for one figure A, and even if the reflected light level is the same, the intermediate level signal (d) changes depending on its position. The comparison signal (E) from the comparator 08 becomes larger or smaller than the level signal (D).
However, it does not correspond to the output (a) from the sample and hold circuit 07, and the pattern of figure A output to the processing unit 5 is deformed compared to the actual figure A, such as the line width becoming thicker or thinner. There was a drawback that it became something like that.

Furthermore, as shown by the lower line in Fig. 3, when scanning a part where the distance between adjacent figures A is large, Fig. 9 (a)
As shown in FIG. 3, the output (C) from the second peak value detection circuit 01G changes greatly, and the shape of the figure A is significantly more deformed than the actual figure A. In order to deal with such a case, it is conceivable to increase the time constant (ct-Rt) of the second peak value detection circuit 010, but conversely, as mentioned above, if the interval between adjacent figures A is As shown in FIG. 9(b), when the output from the second peak value detection circuit O1O is
Depending on how the intermediate level signal (D) is set, there may be cases where it is determined that there is no contact with the figure, especially when the printing is light. However, it is practically impossible to set a suitable intermediate level signal (D), and it is impossible to set a comparison signal (E) that is faithful to the actual figure A.
There was a drawback that it was difficult to obtain.

(Objective of the Invention) The present invention has been made in view of the above circumstances, and it is possible to easily set an intermediate level signal, but it is also possible to easily set an intermediate level signal that is faithful to the actual figure, regardless of the arrangement pattern of the figure. A comparison signal is obtained to perform figure recognition processing well, and a plurality of intermediate level signals are used to obtain a plurality of comparison signals, and the comparison signals and standard figure patterns are ranked for matching judgment. The object of the present invention is to enable good and quick figure recognition without being affected by the difference in shading of printed characters.

(Structure and Effects of the Invention) In order to achieve the above object, the figure recognition processing device of the present invention uses the first reflected light from the recorded portion of the figure recorded on the recording medium and the area other than the line recorded portion. a lens system that converges the second reflected light from the second reflected light to form an image;
24th photoelectric conversion means for photoelectrically converting the reflected light and outputting it as a first level signal and a second level signal, respectively according to the level of the reflected light; 1. clamping means for clamping and outputting the second level signals, peak value detection means for detecting the peak value of the first level signal clamped and outputted from the clamping means and outputting an envelope; a setting means for setting and individually outputting a plurality of intermediate level signals having a level between this and the second level signal and different from each other based on the peak value detection output from the means; and the clamping means. a comparison means for individually comparing the first level signal from the plurality of intermediate level signals with each of the plurality of intermediate level signals and outputting a plurality of comparison signals individually corresponding to each intermediate level signal; and a comparison signal from the comparison means. Among these, calculating the degree of similarity of the recorded figure pattern on the recording medium to the standard figure pattern based on the intermediate level signal of the level signal of the first level signal, and determining whether or not the recorded figure pattern can be recognized as a result of this calculation. When it is determined that the recorded graphic pattern can be recognized through the processing using the comparison signal corresponding to the intermediate level signal located at the level of the first level signal among the respective intermediate level signals. and a recognition processing unit that performs recognition processing of the recorded graphic pattern.

In other words, the first and second level signals output from the photoelectric conversion means are clamped to eliminate fluctuations in the output level caused by the lens system, and to eliminate fluctuations in the output level caused by differences in printing shading among both level signals. The peak value of only the first level signal, which is likely to fluctuate, is detected and the envelope output is performed. Level signals are obtained, digital comparison signals are individually output using these intermediate level signals, and among the plurality of comparison signals, the standard The degree of similarity with the figure pattern is calculated, and it is determined based on the calculation result whether or not it is recognizable. When it is determined that the figure is recognizable, recognition processing is performed to recognize the figure.

Therefore, even if the signal level output from the photoelectric conversion means fluctuates due to a decrease in the amount of light at the periphery of the recording medium due to the lens system, it is stabilized by the clamp, so the level is output as a second level signal. Although the first level signal has level fluctuations due to printing shading, etc., by making the envelope output one with little level change, it is possible to obtain a signal with little variation. By using the constant second level signal, the intermediate level signal can be set extremely easily, and a comparison signal faithful to the actual figure can be obtained in good condition regardless of the arrangement pattern of the figure. Ta.

Furthermore, by taking advantage of the fact that the intermediate level signal can be easily set and setting multiple intermediate level signals that are different from each other, it is possible to avoid problems such as printing that is too light (that is, it is determined that there is no figure), and A comparison signal obtained from a level signal is subjected to a matching judgment with a standard figure pattern in order from a comparison signal based on an intermediate level signal closer to the first level signal, and the judgment is terminated when it is judged that it can be recognized. This eliminates the need for unnecessary judgments, increases the judgment processing speed, and overall allows shapes to be recognized without being affected by differences in printing shading, as well as to quickly perform matching judgments. It has become possible to perform figure recognition processing well and quickly.

(Description of Examples) Hereinafter, the present invention will be described in detail based on examples shown in the drawings.

FIG. 2 shows a schematic configuration of a figure reading/recognition device according to an embodiment of the present invention, and its operation is as described above.

FIG. 1 is a circuit diagram showing details of the amplification/A/D conversion section 4 and the recognition processing section 5 in FIG. 1st and 2nd
It is equipped with an amplifier 6 for amplifying the level signals S , , S , a sample-and-hold circuit 7 , a clamping means 8 for clamping the sample-and-holding output of the sample-and-holding circuit 7 , a peak value detecting means 9 , a setting means 1O, and a comparing means 11 . ing.

The clamping means 8 includes a capacitor 12 and a resistor 13 (time constant C1/R) for removing low-frequency components of the output of the photoelectric conversion means 3.
1) It consists of a diode 14 and a buffer amplifier 15 that keep the level of the first level signal, which is a bright level, at a constant voltage (+V*), and the input first and second level signals
The level signals S, , S are each configured to be clamped by cutting off a fluctuation component of the output of the photoelectric conversion means 3 caused by the lens system 2. Note that a plurality of clamping means 8 may be provided as shown by broken lines 8a in FIG.

The output from the clamp means 8 is input to the peak value detection means 9. This peak value detection means 9 is an operational amplifier! 6,
Diode 17, capacitor 18, resistor 19 (time constant 0
4.R,) and a buffer amplifier 20,
Moreover, by inputting the gate signal g to the operational amplifier 16 and setting the time constant 04·R4 to a large value, either the upper or lower line L1. Even when scanning with L t, that is, regardless of the arrangement pattern of figure A..., as shown in FIG. 4 (b) and (d), respectively,
The configuration is such that fluctuations in the envelope output S3 of the second level signal S can be reduced while keeping the level of the second level signal S constant.

The output from the peak value detection means 9 is input to the setting means IO, and as shown in FIG. 4(b), the envelope output S of the second level signal SI and the second level signal S are A plurality of intermediate level signals S a, S b.

Comparators 11a, lb, 11 which configure the comparing means 11 set the intermediate level signals Sa, Sb, and Sc.
Each C is output individually.

Then, each of the comparators 11a, llb, and llc has the following:
The output from the clamping means 8 is input as is, and by comparing the magnitude with the intermediate level signals Sa, Sb, Sc, a plurality of comparison signals corresponding to the first and second level signals S, S, are outputted. ing.

As shown in FIG. 1, the recognition processing unit 5 includes a comparator 11a,
, 1 lb, and 11c, respectively, an encoder 21 that encodes and outputs the comparison signals, an image memory 22 as a storage means for storing the encoded outputs CO and Ql from the encoder 21, and a control/judgment section 23. It consists of

The image memory 22 includes a first memory 22a that stores one of the encoded outputs GO and CI from the encoder 21, C0;
On the other hand, it has a second memory 22b that stores C1.

The control/judgment unit 23 compares the coded outputs co, ct from the image memory 22 with comparison outputs c j, c l,
A comparator 24 converts the coded outputs cO and cI into n-values through this comparison and outputs them, and a projection length counter that counts the n-value outputs from the comparator 24 and measures the projection length of the recorded figure pattern. 25, a standard figure pattern output section 26 which outputs an n-valued output of a standard figure pattern in response to a drive signal, and a standard figure pattern output section 26 which outputs both n-valued outputs, and generates a recorded figure pattern based on both n-valued outputs. A superposition circuit 27 outputs a superposition signal indicating the state of superposition with the standard figure pattern, and supplies the drive signal to the standard figure pattern output section 26, and outputs a measurement output from the projection length counter 25 and a superposition signal from the superposition circuit 27. It has a processing circuit 28 for processing the superimposed output and the second level signal S! with respect to the comparison signal from the comparison means 11. Based on the intermediate level signal of the level signal of (S a-* S
b-+ S c), while calculating the similarity of the storage graphic pattern of the recording medium l with respect to the standard graphic pattern,
As a result of the calculation, it is determined whether or not the recorded figure pattern can be recognized, and when it is determined that the recorded figure pattern can be recognized as a result of the above calculation and judgment, the conditions of the recorded figure pattern are determined. It is configured to perform recognition processing.

Next, the operation of the above recognition process will be explained using the flowchart shown in FIG.

First, a first comparison output (C
' ←1. In step Nl), the image memory 22 corresponds to the intermediate level signal Sa located at the first level closest to the second level signal S1 among the intermediate level signals Sa, Sb, and SC. The recorded figure pattern based on the comparison signal is input to the projection length counter 25 and the overlay circuit 27, the output from the standard figure pattern output section 26 is input to the overlay circuit 27, and each figure on the recording medium 1 is cut out. , the corresponding recorded graphic pattern and the standard graphic pattern are compared to calculate the degree of similarity (step N2).

Next, as a result of the similarity calculation, it is determined whether the recorded graphic pattern can be recognized (step N3), and if it is determined that it is recognizable, recognition processing is performed (step N4). On the other hand, if it is determined that it cannot be recognized, the processing circuit 28 sends a signal to the comparator 24.
The second comparison output (C0-o, c'←1) is inputted to step N5), and the intermediate level signal S is input from the image memory 22.
an intermediate first level signal S among a, S b, and S c;
intermediate level signal s located at the second closest level from 1
The recorded graphic pattern based on the comparison signal corresponding to b is input to the projection length counter 25 and the superposition circuit 27, and the degree of similarity of the recorded graphic pattern with the standard graphic pattern is calculated in the same manner as step N2 (step N6). after that,
It is determined whether the recorded graphic pattern can be recognized from the result of the similarity calculation (step N7), and if it is determined that it is recognizable, recognition processing is performed in step N4. If it is determined that it cannot be recognized, the -processing circuit 28 sends a third signal to the comparator 24.
Step N to input the comparison output (cO←1.Cl-0)
8), intermediate level signals Sa, Sb from the image memory 22
, Sc is input to the projection length counter 25 and the superimposition circuit 27, and the recorded figure pattern is inputted to the projection length counter 25 and the superimposition circuit 27. N6
Similarity between the recorded graphic pattern and the standard graphic pattern is calculated in the same manner as (step N9).

Thereafter, it is determined whether the recorded graphic pattern can be recognized from the result of the similarity calculation (step N10), and if it is determined that it is recognizable, recognition processing is performed in step N4. If it is determined that it cannot be recognized, it is finally processed as unrecognizable (step N11).

Next, the effects of the above embodiment will be explained.

That is, based on the intermediate level signals S a, S b, and S c, comparison signals D+ and Dt individually corresponding to each one are generated.
, D, are input to the encoder 21 as shown in FIG. 4(C), and the encoder 21 outputs a coded output C1.
, C proverbs are input into the image memory 22. And the sixth
A number r with a difference in shading in the left and right directions, as shown in figure (a)
In a case like OJ, first, 1 from the first level signal.
According to the comparison signal corresponding to the intermediate level signal Sa, as shown in FIG. 6(b), the line width as a whole becomes thinner, and the thin areas of printing are not extracted as recorded figure patterns, resulting in recognition. judged to be impossible. Then,
It is determined whether the comparison signal I corresponding to the second intermediate level signal sb from the first level signal can be recognized. In this case, as shown in FIG. 6(C), although there is a change in line width, ” and is determined to be recognizable here.
Accompanying this, recognition processing is performed and the figure cut out from the recorded figure pattern is recognized as "0".

According to the present invention, in the above-mentioned case, the recognition process is terminated at this point, and no determination is made as to whether or not recognition is possible based on the comparison signal corresponding to the third intermediate level signal Sc from the first level signal. Since the process automatically moves on to the recognition process for the next figure, the speed of the recognition process for the figure can be increased.

This is because, in the case described above, even if it is determined whether recognition is possible using the comparison signal corresponding to the third intermediate level signal Sc from the first level signal, the result shown in FIG.
As shown in d), the line width becomes thicker, and as a result of superimposition with the standard figure pattern, the number of mismatched pixels increases and becomes unrecognizable. For example, there are always multiple intermediate level signals Sa,
Even if recognition determination is performed using comparison signals corresponding to both Sb and Se, the recognition processing speed will only become slower.

With the present invention, it has become possible to perform figure recognition satisfactorily in the minimum amount of time necessary.

In the above embodiment, a blank paper is used as the recording medium l, on which figures A, such as numbers and letters, are printed, that is, a first level signal S1 which is the reflected light level of the recording part.
is lower than the second level signal S, but in the present invention, for example, black paper is used as the recording medium l, and the figure A... is printed therein using a high brightness color such as white. The present invention can also be applied to the case where the first level signal S1 is higher than the second level signal S.

The present invention is not limited to card-based recording media such as betting tickets such as horse racing tickets and car tickets, but can also be applied to recording media such as documents, for example.

[Brief explanation of the drawing]

FIG. 1 is a circuit diagram showing an embodiment of the figure recognition device of the present invention, FIG. 2 is a schematic configuration diagram of the figure recognition device, FIG. 3 is a front view showing an example of a recording medium, and FIG. 4(a) , (b), (c)
and (d) are waveform diagrams of signals and outputs, FIG. 5 is a flowchart showing the operation in the recognition processing section, and FIG. 6(a)
is a diagram showing the correlation between the figure and the intermediate level signal, FIG.
), (c) and (d) are diagrams showing output graphic patterns corresponding to the first to third intermediate level signals, respectively;
Figure 7 is a circuit diagram of the conventional example, Figures 8 and 9 (a)
, (b) is a waveform diagram of signals and outputs of a conventional example. l... Recording medium, 2... Lens system, 3... Photoelectric conversion means, 5... Recognition processing section, 8... Clamping means, 9
...Peak value detection means, IO...setting means, 11.
... Comparison means, 21.. Encoder, 22.. Storage means, 23.. Control/judgment section, 24.. Comparator, 25.
...Projection length counter, 26.. Standard figure pattern output section, 27.. Superposition circuit, 28.. Processing circuit.

Claims (3)

[Claims] (1) A lens system that converges a first reflected light from a recorded portion of a figure recorded on a recording medium and a second reflected light from a place other than the recorded portion to form an image; the first and second
a photoelectric conversion means for photoelectrically converting the reflected light and outputting it as a first level signal and a second level signal, respectively according to the level of the reflected light; and a first, clamping means for clamping and outputting the second level signals, peak value detection means for detecting the peak value of the first level signal clamped and outputted from the clamping means and outputting an envelope; and the peak value detection. a setting means for setting and individually outputting a plurality of intermediate level signals having a level between this and the second level signal and different from each other based on the peak value detection output from the means; and the clamping means. 1st from
Comparing means for individually comparing the level signal with each of the plurality of intermediate level signals and outputting a plurality of comparison signals individually corresponding to each of the intermediate level signals; Calculating the degree of similarity of the recorded graphic pattern on the recording medium to the standard graphic pattern based on the intermediate level signal that is closer to the level of the 1-level signal, and determining whether the recorded graphic pattern can be recognized as a result of this calculation. When the processing is started and it is determined that the recorded graphic pattern can be recognized by the processing using the comparison signal corresponding to the intermediate level signal located at the n-th level from the level of the first level signal among the respective intermediate level signals. A figure recognition processing device comprising: a recognition processing unit that performs recognition processing of the recorded figure pattern; (2) In the figure recognition processing device according to claim 1, the recognition processing unit includes an encoder that encodes and outputs each comparison signal from the comparison means, and a code from the encoder. the encoded output stored in the storage means; and a function of comparing the recorded graphic pattern on the recording medium with a standard graphic pattern to recognize the recorded graphic pattern, and storing the encoded output stored in the storage means. A figure recognition processing device comprising: a control/judgment unit that controls the calculation and determination processing based on the above. (3) In the figure recognition processing device according to claim 2, the control/judgment unit compares the coded output from the storage means with a comparison output, and based on this comparison, the coded output a comparator that outputs the n-value output from the comparator; a projection length counter that measures the projection length of the recorded graphic pattern by counting the n-value output from the comparator; a standard figure pattern output section that outputs converted output;
a superimposition circuit which is provided with both n-valued outputs and outputs a superposition signal indicating the superimposed state of the recorded figure pattern and the standard figure pattern based on both the n-valued outputs; and a superimposition circuit which outputs the drive signal into a standard figure pattern. In addition to giving to the department,
A figure recognition processing device that includes a processing circuit that processes measurement output from a projection length counter and superposition output from a superposition circuit.