JPS62171083A - Optical type character reading method - Google Patents

Abstract

PURPOSE:To absorb the deviation of the focus of image data by adding the value to multiply a negative constant to the value of the two-step difference of the image data after A/D conversion to original image data and shaping the waveform. CONSTITUTION:The output waveform of an A/D converting circuit 5 is inputted to the first step latch circuit 6-3 and the output an+1 is inputted to a ROM circuit 6-6 and simultaneously, inputted to the second step latch circuit 6-4. The output an of the latch circuit 6-4 is inputted to the ROM circuit 6-6 and simultaneously, inputted to the third step latch circuit 6-5. The output an-1 of the latch circuit 6-5 is inputted to the ROM circuit 6-6. In the ROM circuit 6-6, the data to calculate three pieces of the data of an+1, an and an-1 inputted to the address terminal in accordance with the formula a'n=an-k(an+1-2an+ an-1) are accommodated beforehand, and data a'n after they are converted by the formula are outputted by an output terminal 6-2.

Description

DETAILED DESCRIPTION OF THE INVENTION (Field of Industrial Application) The present invention relates to an optical character reading method, and particularly to waveform shaping of image data obtained by photoelectric conversion.

(Prior Art) FIG. 6 shows a general configuration of a conventional optical character reading device (hereinafter referred to as OCR). In Figure 6, 21 is a form,
The image on this form 21 is illuminated by a light source 22, and the reflected light is passed through an imaging lens 23 to a CCD sensor 2.
4 receives. The CCD sensor 24 converts the received optical signal into an electrical signal, and the A/D conversion circuit 25 digitizes this electrical signal and stores it in the image memory 26. On the other hand, the two-dimensional image data on the form 21 is stored in the image memory 26 by the transport mechanism 27 transporting the form 21 in the direction of the arrow 28 .

The contents of the image memory 26 are then processed by a recognition unit (not shown), but in this case, it is important to accurately reproduce the image on the form 21 in the image memory 26.

(Problems to be Solved by the Invention) However, in the above-mentioned conventional optical character reading method in OCI+, errors in adjustment of the imaging lens 23 in the direction of the arrow 29 and ups and downs in the direction of the arrow 30 of the form 21 cause C
There is a problem in that the image reflected on the CD sensor 24 is out of focus. An out-of-focus waveform has a distorted shape as shown in Figure 7. FIG. 7 partially shows the output waveform of the CCD sensor 24, where ■ is the voltage, t is the time,
The solid line shows the corrupted input waveform, and the dotted line shows the ideal input waveform without the corruption. In Figure 7,
Pure white on the form 21 corresponds to VW on the V axis, and pure black corresponds to VB.

Compared to the ideal waveform shown by the dotted line, the actual waveform is usually rounded as shown by the solid line.

The present invention was made in order to solve the problems of the prior art, and an object of the present invention is to provide an optical character reading method that can improve waveform distortion and obtain correct image data. do.

(Means for Solving the Problems) The present invention optically scans an image on a form to obtain an optical signal of the image, converts the obtained optical signal into an electrical signal by photoelectric conversion, and further converts the obtained optical signal into an electrical signal. The present invention is directed to an optical character reading method in which an image on a form is read by performing A/D conversion on the image data to obtain image data, and is among the conventional techniques described above.

In order to solve the illumination point, the second-order difference of the image data after A/D conversion is calculated for each sample time, and the value obtained by multiplying the value of the second-order difference by a predetermined negative constant is calculated for each sample time. Adding to the value of the original image data at′
Accordingly, the waveform of image data is shaped.

(Function) In the optical character reading method of the present invention, the value of the second-order difference of the image data after A/D conversion is multiplied by a negative constant, and the value is added to the value of the original image data. I'm trying to get some plastic surgery done. The former value added to the value of the original image data acts to absorb out-of-focus of the image data due to adjustment errors of the imaging lens, ups and downs of the form, and the like. Therefore, image data ' obtained after waveform shaping has improved waveform distortion, and more accurate data can be provided.

(Example) An embodiment of the present invention will be described below.

Figure 1 shows an OC to which the optical character reading method of the present invention is applied.
It is a figure showing the composition of R. In the figure, 1 is a form, 2 is a light source, and 3
4 is an imaging lens, 4 is a C'CD sensor, 5 is an A/D conversion circuit, 6 is a shaping circuit, 7 is an image memory, and 8 is a transport mechanism for the form 1. In FIG. 1, each element other than the shaping circuit 6 is the same as that of the conventional OCR shown in FIG. Also,
The process up to inputting the image of the form 1 into the A/D conversion circuit 5 is the same as the OCR case shown in FIG. The difference is that it is stored in The shaping circuit 6 includes an input terminal 6-1, an output terminal 6-2, and three latch circuits 6-3.
It consists of a 6-4 degree 6-5 and a ROM circuit 6-6.

Now let's talk about the principle of shaping a corrupted waveform.
When the voltage of the waveform is V, waveform shaping is performed by adding a constant K (K>O) d to the second-order differential of the voltage V with respect to time -d''V-. Fig. 2 <a> ~(
The situation is illustrated in d). In Fig. 2 (a) to (d), V
is the voltage, t is the time, the solid line on the graph is an example of the waveform, point A is one point on the waveform, and the arrow indicates - at point A in Figure 2 (a) ~
Four types of curves as shown in (d) are included, and shaping can be performed by changing each curve in the direction of the arrow.

In the present invention, this shaping is performed on data after A/D conversion. Process the data after A/D conversion.

az+aa+...lat++..., where n is the sampling time and alo is the waveform after shaping, a'n=an-K(ane, 2 a n0a,, -1
) ”' (1) is used. In equation (1), the first term a on the right side.

represents the original waveform, and the second term −K representing the second order difference is added to this
Add (a, , 2 a n + a n-1). If we transform this second term, we get:

K ('a q+12 a, + a n-4)=-
K((an,, -an) (a,, a, -1))
-(2). The right side of equation (2) (a,,.1 an
) is the waveform a at sample time n. There is a difference that represents the size that is going to increase in the future, and (a, an-t)
is the difference in the past of an. Therefore, the second term in equation (1) is -K
It has been multiplied by .

In the OCR shown in FIG. 1, the processing of equation (1) above is realized by the shaping circuit 6. The output waveform of the A/D conversion circuit 5, that is, the input waveform of the shaping circuit 6, is input to the first stage latch circuit 6-3, and its output a. . 1 is input to the ROM circuit 6-6 and simultaneously input to the second stage latch circuit 6-4. Output a of latch circuit 6-4. is input to the ROM circuit 6-6 and simultaneously input to the third stage latch circuit 6-5. Outputs a, -4 of the latch circuit 6-5 are input to the ROM circuit 6-6. The ROM circuit 6-6 has a nil la IT J a input to the address terminal.
The data calculated from the three data n-1 according to formula (1) is stored in advance, and the data calculated according to formula (1) is output from output terminal 6-2.
The converted data aTo is output by.

Next, a specific example of the shaping circuit 6 and its operation will be explained. Third
The figure shows the waveform of FIG. 7 A/D-converted into 16 gradations, where n is the sampling time and a, , is the data after A/D conversion. Figure 3 a. When listed, a□=15. a
2=15. a,=14. a4=12. a, = 9
+a G= 5 y a t = 2 r a m
= Or ag = Or a 1a = 0.

Here, it is assumed that the ROM circuit 6-6 in FIG. 1 stores in advance data specifically calculated by the following equation.

"n" an+t+3a, all-l """
(3) Equation (3) is obtained by setting the constant K in Equation (1) to 1. However, ag as a result of calculation using equation (3). is less than 0, it is O, and when +a'l'l is greater than 15, it is 15. This is because arl is data with 16 gradations. Furthermore, since all has 16 gradations,
The latch circuits 6-3, 6-4, and 6-5 are all 4 bits, and the ROM circuit 6-6 has a 12-bit input and a 4-bit output.
Use bits.

Now, following the example above, a in FIG. FIG. 4 shows how the is processed. "Cycle" in Figure 4
is the sequence number of the tarok input to the latch circuit 6-3.6-4.6-5. Further, the symbol - in FIG. 4 represents the data net. The contents of FIG. 4 will be explained step by step below. In the 0th cycle, the first data a 1=
15 is input to the latch circuit 6-3, and therefore ao. □
=15. In the first cycle, a 1= 1
5 is input to the latch circuit 6-4, and a2=15 is input to the latch circuit 6-3. Therefore a... □” '5
+ a n = 15. In the second cycle, a=15 is input to the latch circuit 6-5, and a2=1
5 is input to the latch circuit 6-4, and a,=14 is input to the latch circuit 6-3. Therefore a,,,□=14゜a
, 1=15. a, l-1=15. This is expressed as formula (3)
When calculated according to the following, a', = 16, but ago is 1
Since it is larger than 5, the output of the ROM circuit 6-6 is 15. If the same processing is carried out thereafter, the ROM circuit 6-6 will appear as shown in FIG. become that way. FIG. 5 shows a+o after shaping. In FIG. 5, the dotted line is the input waveform of the shaping circuit 6, and the solid line is the output of the ROM circuits 6 to 6, and it can be seen that the roundness of the waveform in the solid line is improved more than in the dotted line.

(Effects of the Invention) As explained in detail above, according to the present invention, an electric signal obtained by photoelectrically converting an optical signal of an image on a form is converted into an A/D converter.
After conversion, waveform shaping is performed by calculating the second-order difference of the data, multiplying that value by a negative constant, and adding the value to the original A/D converted data. It is possible to reduce data distortion caused by adjustment errors, fluctuations in forms, etc. Therefore, it can be expected that the OCR will have a higher ability to detect minute loops on documents and thinly drawn lines.

[Brief explanation of drawings]

Figure 1 shows an OC to which the optical character reading method of the present invention is applied.
Figures 2 (a) to (d) are diagrams showing examples of corrupted waveforms, Figure 3 is a diagram showing waveforms A/D converted to 16 gradations, and Figure 4 is a diagram showing the configuration of R. Figure 3 shows how the data is processed, Figure 5 shows the waveform after shaping, Figure 6 shows the configuration of a conventional OCR, and Figure 7 shows the CCD sensor of the OCR in Figure 6. It is a figure which shows a part of output waveform. 1... Form, 2... Light source, 3... Imaging lens, 4... CCD sensor. 5... A/D conversion circuit, 6... Shaping circuit, 6-
3.6-4.6-5...Latch circuit. 6-6...ROM circuit, 7...image memory, 8...transport mechanism.

Claims (1)

[Claims] The image on the form is optically scanned to obtain an optical signal of the image, the obtained optical signal is converted into an electrical signal by photoelectric conversion, and the electrical signal is further subjected to A/D conversion. In an optical character reading method that reads the image on a form by obtaining image data, the second-order difference of the image data after A/D conversion is calculated for each sample time, and the value of the second-order difference is calculated. An optical character reading method characterized in that waveform shaping of image data is performed by adding a value multiplied by a predetermined negative constant to the value of the original image data at each sample time.