JPS62200485A - Optical character reader - Google Patents

Abstract

PURPOSE:To read a patchy character written in a slip by separating an optical signal from a reading slip to a color component and prohibing to generate reading image data when the light signal is decided to be a dropping-out color. CONSTITUTION:At the paper surface of an ordinary slip and the part where the character is written or printed in the paper surface, a wavelength reflecting ratio curve is approximately flat. Consequently, concerning these parts, a transmitting light quantity per unit wavelength width of respective color filters (R, G and B) to receive the reflection of a white light from the slip is approximately the same value. On the other hand, at the part where respective red type, green type and blue type dropping-out colors are pre-printed to an OCR reading slip, the wavelength reflecting ratio curve of the dropping-out color of respective color systems comes to be the inherent one to have a peak with a special wavelength. Consequently, it can be decided from the transmitting light quantity ratio of a color filter whether or not the part is the part of the dropping-out color and at the part of the dropping-out color, the image cannot be picked up.

Description

[Detailed description of the invention] [Purpose of the invention] (Industrial application field) The present invention relates to an optical character reading device.

(Prior Art) In recent years, with the spread of optical character reading devices (hereinafter referred to as "OCR"), there is a strong demand for the realization of OCR that can read extremely low quality characters that were conventionally considered unreadable. Such low-quality characters include noisy characters, such as characters that are smudged by the printer's ribbon being rubbed or ink splattered from the printer head; Characters that are entirely or partially faded, such as printed characters or characters written with a ballpoint pen (hereinafter referred to as low PC8 characters)
There is. For the former type of noisy characters, countermeasures such as noise removal using firmware or the like can be considered, but at present there is no effective countermeasure for the latter type of low PC8 characters. In the case of low PC8 characters, part or the entire signal component of the character is already lost at the stage of the photoelectric conversion system that detects the character, and the lost character information is almost never recovered at the stage after the optical conversion system. Because it's impossible.

2 of the detection optical system to improve the reading rate of low PC8 characters.
It may be possible to adjust the value detection threshold to detect characters with low Pcs, but with conventional photoelectric conversion systems, if the threshold is lowered, characters printed in dropout colors on forms can be detected. This also results in the detection of eta in the entry frame, etc., and conversely increases the error rate in character reading.

(Problems to be Solved by the Invention) As described above, the conventional OCR has a problem in that it is not possible to read eight low PC characters without error by adjusting the binarization threshold of photoelectric conversion.

The present invention was made to solve the above problems, and
To provide an optical character reading device capable of reading even low PC8 characters written on a form preprinted with a dropout color.

(Means for Solving the Problems) The present invention provides an optical system that photoelectrically converts an optical signal from a read document, quantizes the photoelectrically converted electric signal, and generates read image data based on the quantized digital signal. The target character reading device includes dropout color determining means for separating the optical signal from the read form into color components and determining whether or not the optical signal is a dropout color based on the separated color components, The present invention is characterized in that generation of the read image data is prohibited when the dropout color determining means determines that the optical signal has a dropout color.

(Function) According to the present invention, since the optical character reading device is configured as described above, even if the threshold value of the photoelectric conversion system is lowered, the top-out color does not appear, and only the characters can be properly read and converted into image data. can do.

(Embodiment) FIG. 1 shows an optical character reader 1 according to an embodiment of the present invention, in which white light is irradiated from a light source (not shown), and the reflected light is filtered through filters having spectral characteristics of red, green, and blue, respectively. 2,3.4
incident on . The reflected light that has passed through the red filter (R) 2 is collected by a lens 5 and sent to a line sensor 8 which is a photoelectric conversion means.
incident on .

The reflected light that has passed through the green filter (G) 3 is also collected by the lens 6 and enters the line sensor 9. Also attached filter (B)
The reflected light that has passed through the line sensor 4 is also collected by the lens 7 and enters the line sensor 10. A clock signal is input from the outside to the line sensors 8, 9.10, and the line sensor °8.9.1
0 outputs a photoelectric conversion output signal of one light receiving element of each line sensor for each pulse of the clock signal. The photoelectric conversion output of the line sensor 8 is output by an amplifier (AMPI '>11
The comparator (CMPl) 14 and the comparator (CMP
2) Supplied to 15 six inputs. The photoelectric conversion output of the line sensor 9 is transmitted through an amplifier (AMP2) 12 to a comparator (C
MP1) 14 and six inputs of a comparator (CMP3) 16. In addition, the photoelectric conversion output of the line sensor 10 is transmitted through an amplifier (AMP3) 13 to a comparator (CMP2) 15.
and is supplied to the B input of the comparator (CMP3) 16. The comparison outputs (C1C2, C3) of the comparators 14, 15, and 16 are input to the dropout color determination circuit 17.

On the other hand, the amplified outputs of amplifiers 11, 12, and 13 are
8 to an amplifier (AMP4) 19. The amplifier 19 outputs the output (P) of the dropout color determination circuit 17.
is input as a control signal, and the amplified output is input to the sample hold circuit (SH) 20. Sample hold circuit 2
The output of 0 is input to the correction circuit 22 via the A/D converter 21, and the corrected signal is supplied to the buffer memory 23.

The recognition unit 24 inputs image data from the buffer memory 23 and outputs candidate categories as a result of character recognition to the outside.

First, the operating principle of this embodiment will be explained. FIG. 2 is a wavelength reflectance curve diagram for explaining the operating principle of this embodiment, in which the horizontal axis represents wavelength and the vertical axis represents reflectance. Figure 2 shows examples of wavelength reflectance curves for pencils used to fill in OCR reading forms, black ballpoint pens, printer ribbons containing graphite, and dropout colors of each white family, as well as spectral characteristics of red, green, and blue. An example of the transmission range of the color filter (R, G, B) (corresponding to the 11th code 2.3.4) is shown. According to this, the wavelength reflectance curve is almost flat on an ordinary paper surface or on a portion where characters are written or printed on the paper surface. Therefore, for these parts, each color filter (R, G
, B) have almost the same amount of transmitted light per unit wavelength width. On the other hand, in the area where red, linear, and front thread dropout colors are printed on the OCR reading form, the wavelength reflectance curve of each white dropout color has a unique peak at a specific wavelength. As a result, color filters (R,
The amount of transmitted light per unit wavelength width of G and B) differs depending on the dropout color of each white system. Therefore,
From the amount of transmitted light per unit wavelength width of the three types of color filters (R, G, , B), the ratio of the transmitted light m of the two types of color filters (for example, the amount of transmitted light of color filter R and the amount of transmitted light of color filter G) Furthermore, the ratio of the amount of transmitted light of two types of color filters in other combinations (for example, the ratio of the amount of transmitted light of color filter R, the amount of transmitted light of color filter B, and the amount of transmitted light of color filter B) is determined. If the ratios of the former and latter are both close to 1, then the object to be detected (i.e., a form) from which the white light was irradiated and the reflected light was obtained is a normal paper surface, or there are characters written on it. It may be considered as a printed or printed part. Moreover, if either one of these ratios is a value far from 1, it is considered to be a dropout color briprint portion. Therefore, when OCR reads characters on a form, if you do not want to pick up the image of the dropout color print area, check whether the dropout color is in the dropout area 1 to color from the transmitted light m ratio of the color filter. This means that it is possible to prevent images from being picked up in the dropout color area. That is, when it is determined that the color is dropout, the amplifier for the image signal is not operated in the saturation region. On the other hand, if it is determined that the color is not a dropout color, the image signal amplifier may be operated in a linear region.

Next, the operation of this embodiment will be explained. The light reflected from the form 1 passes through a color filter (R, G, B) 2°3.4, and as shown in Figure 2, the light passes through a red transmission region, a green transmission region,
The transmitted light corresponds to the blue transmission region and enters the line sensor 8.9.10 via the lenses 5, 6.7, respectively. Amplifiers 11, 12, and 13 amplify the photoelectric detection power signals of line sensors 8, 9, and 10, and input amplified signals corresponding to the amount of light transmitted through each color filter to comparators 14, 15, and 16. Here, the comparisons 314, 15, and 16 are the logical outputs (C1, C2, C3
), and the operation of the comparator will be explained below with reference to FIG. The third figure shows the comparator output C (outlet C1 of each comparator,
(corresponding to C2 and C3). In other words, the position of the point determined by the input to and the input to B is the third
In the area indicated by diagonal lines in the figure, the comparator generates a "1" output at the C output, indicating that there is a significant difference between the A and 8 inputs; When the position of the determined point is outside the shaded area in Figure 3, the comparator generates °O'' at the C output, indicating that there is no significant difference.
The case when O is shown in equation (1), and the case when C=1 is shown in equation (2).

C=O((AHA,,,1) (BI8..1))
((A>An, □)B (B>BTl12)) (<<α)
(1) α A C-1 ((ATI11≦Δ≦AT112) (B1o1≦
B≦BTI+2)) The above formula is input to or threshold value Arl+1
is within the range of the threshold value Δ1112, and the B input is 8□11
1 to the threshold" Br112, and when the ratio of the 8 input to the input is smaller or larger than 1/α, the output C becomes 1", otherwise OII
It shows that. Here, threshold A, threshold ■1
11 value BTi11 indicates the lower limit of the inputs and 8 inputs, respectively, and if the Δ input or the B input has a value smaller than the above VA value, it is determined that the color is extremely close to black. −h, threshold (+CiA
, lal value B is input to each, 8 input ■++1
TH2 is the upper limit, and if the input to or B input has a value larger than the above threshold value, it is determined that the color is extremely close to white.

Also, α is a value that defines the upper and lower limits of the ratio B/Δ between input to input and input B, and if the ratio B/A is within the range of 1/α to α (however, α≧1), input to input and 8 It is determined that there is no significant difference between the input and the input.

In the embodiment shown in FIG. 1, the outputs C (C1°C2, C3) of the comparators 14, 15, and 16 that perform the above judgment logic operation
is input to the dropout color determination circuit 17. The dropout color determination circuit 17 includes comparators 14 and 15.1.
The overall judgment as described below is made from the output of step 6, and when it is judged as a dropout color, 1" is output on the output, and when it is judged as not a dropout color, "0" is output on the output. Figure 4 shows the dropout color judgment. A specific example of the circuit 17 is shown. Outputs of comparisons 514, 15, and 16 (C, C, C3
) are each inverter 171 with Schmitt trigger input.
, 172, 173 through 3-man NANO gate 17
Enter 4.

The output of the NAND gate 174 is the output of the dropout color determination circuit 17. According to the above configuration, one of the outputs C1, C2, and C3 of the comparator is “1”.
'', the output of the NANO gate 174 will be 1''. As a result, if even one of the two combinations of the three types of transmitted light amounts of the reflected light transmitted through the color filters (R, G, B) is far from 1, the comparator 14°15.16> Since the C output of any one of is 1'', N
The output of the AND gate 174 becomes "1", and it is determined overall that it is a dropout color part.

The D signal, which is the result of the overall determination by the dropout color determination circuit 17, is supplied as a control signal to the amplifier 19 in the embodiment of FIG. Now, the first
In the illustrated embodiment, the amplified outputs of the amplifiers 11, 12, and 13 are summed via an adder circuit (ADD) 18, and the adder 7 output is input to an amplifier 19. The operating state of the amplifier 19 is controlled by the D signal of the dropout color determination circuit 17, and operates in a saturated state when the D signal is 1'' indicating a dropout color.On the other hand,
“0” indicating that the D signal is not a dropout color
When , it operates in a straight line.

Therefore, when it is determined that the color is a dropout color, the amplifier 19 supplies the sample and hold circuit 20 with an output when the paper surface of the form is close to absolute white, regardless of the output of the adder 18. On the other hand, if it is not determined that the color is a dropout color, the amplifier 19 supplies an output proportional to the human power from the adder 1 to the sample and hold circuit 20.

The output signal of the amplifier 19 obtained in this manner corresponds to a so-called image signal output from a photoelectric conversion section in the prior art. The sample and hold circuit 20 temporarily holds information of the image signal using a clock signal. NaJ
3. The above clock signal is a clock signal synchronized with the driving clock inputted to the line sensors 8, 9, and 10. The image signal held in the sample hold circuit 20 is A/D
It is converted into a digital signal via a converter 21 and input to a correction circuit 22. The correction circuit 22 performs sensitivity correction, shading correction, etc., and sends the corrected signal to the buffer 1 memory 23.
Store in. The recognition unit 24 reads image data from the buffer 7 memory 23, performs character detection, cutting, feature extraction, recognition processing, etc. using known techniques, and outputs candidate categories to the outside as a reading result.

Note that the present invention is not limited to the above embodiments, and various modifications are possible. For example, the dropout color determination circuit may be configured by a circuit different from the above circuit.

〔Effect of the invention〕

As described above, according to the present invention, reflected light from a form is divided into color components and detected, and based on the ratio of the color components, it is comprehensively judged whether or not the dropout color is 1 to 1 part. Since the image is not picked up in some areas, it is possible to read image data of characters with extremely low PC8.

[Brief explanation of drawings]

FIG. 1 is a block diagram of an optical character reading device according to an embodiment of the present invention, FIG. 2 is an explanatory diagram for explaining the operating principle of the optical character reading device, and FIG. 3 is a block diagram of the optical character reading device according to an embodiment of the present invention. FIG. 4 is a circuit diagram showing an example of the configuration of a dropout color determination circuit in the optical character reading device. 1... Form, 2, 3.4... Color filter, 5.6.
7...Lens, 8,9.10...Line sensor, 1
1゜12.13...Amplifier, 14,15.16...
Comparator, 17...dropout color determination circuit, 1
8...Addition circuit, 19...Amplifier, 20...Sample hold circuit, 21...A/D converter, 22...
・Correction circuit, 23...Buffer memory, 24...Recognition unit Fig. 1

Claims (1)

[Claims] 1. In an optical character reading device that photoelectrically converts an optical signal from a read form, quantizes the photoelectrically converted electric signal, and generates read image data based on the quantized digital signal, dropout color determining means for separating an optical signal into color components and determining whether or not the optical signal is a dropout color based on the separated color components; An optical character reading device characterized in that generation of the read image data is prohibited when it is determined that the read image data is out-color.