JPS59160273A - Character reader - Google Patents

Abstract

PURPOSE:To execute a key input of a corrected data by displaying a character data and a business form data by a binary code and a variable density picture, respectively. CONSTITUTION:A character group of a read business form 7 is scanned by a raster scan and photoelectrically transduced in a photoelectric converting part, and binary-coded. Subsequently, the picture signal of one character portion of said obtained picture signal is cut and sent to a read part 2. In this read part 2, a character recognition is executed, and its character code is stored in a read result buffer 3. On the other hand, output of the photoelectric transducing part 1 is sent to a density varying circuit 8, too. In this density varying circuit 8, an (n)X(n) picture element is fetched from in an input picture data arrayed two-dimensionally, the density value of one picture element portion corresponding to said picture element, and a data compression is executed.

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention relates to a character reading device such as a kanji OCR, and in particular, it is capable of easily correcting when unreadability or misreading occurs in the character recognition. It relates to a character reading device.

[Technical background of the invention and its problems 3 A paper character reading device scans characters printed or handwritten on the ζ surface using methods such as photoelectric conversion, analyzes and recognizes the obtained image data. This is a device that determines what kind of character it is and generates a corresponding character code. FIG. 1 is a block diagram of the hardware of this conventional example.

In FIG. 1, a group of characters on a form to be read are photoelectrically converted and binarized by a photoelectric converter (1). Then, an image signal for one character is cut out from this image signal and sent to the reading section (2). Character recognition is performed here, and the results are stored in a reading result buffer (3). This output is sent to the display (4) and displayed on the screen. However, in this type of device, unreadable characters or erroneous readings occur due to dirt on the form, poor print quality, irregular handwritten font, etc. In such cases, the operator normally operates the controller (
A correct character code is entered by key operations on the keyboard (6) connected to the computer 5), and the previously generated character code is replaced with this. However, this checking method has the disadvantage of poor work efficiency.

Conventionally, a display device has been proposed in which the character group ``--'' is directly or indirectly displayed on the display means in parallel with the read information from the recognition means (Japanese Patent Laid-Open No. 56-114
076). According to this method, it is also possible to display multiple pieces of read information, and checking can be performed effectively. However, in this method, the 2
The image data of the value pattern is displayed as is on the display, and is (512 dots) x (512 dots)
When displaying on a standard type display, if the original data is 8 pits 1, only 6.4 (7) x 6.4 cIn original data can be displayed.

In other words, this method has the disadvantage that only a small amount of image data of the input character group is displayed. However, it is inconvenient when making long contextual corrections that involve a lot of kanji. Furthermore, in order to display more of the original data of this input character group, the displayed characters can be made smaller, but if the pixel density is reduced by thinning out the binary pattern, the original data cannot be faithfully reproduced.

In other words, when displayed in a binary pattern, the outline of the character becomes a rectangle and is not smooth, so if this is thinned out, the shape will be different from the original data and it will be difficult to distinguish it with the naked eye. In particular, there is a problem with Chinese texts, etc., and it is difficult for the operator to correct them.

[Purpose of the invention]

An object of the present invention is to provide a character reading device that simultaneously displays the result of character reading and original data on a display, and allows an operator to easily make corrections while referring to a large amount of original data that can be easily identified with the naked eye. Our goal is to provide the following.

[Summary of the invention]

The present invention reads an image signal of a group of input characters to be read, obtains its character data, and obtains compressed and changed gray-scale image data by applying predetermined gray-scale weighting to the image signal of the input character group. The character pattern corresponding to the character code of the reading result and the grayscale image data of the form are displayed simultaneously on the display of the character reading device. That is, in the present invention, character data such as reading results are displayed in binary form as in the conventional system, but form data is displayed in gray scale images. Therefore, by referring to both of them at the same time, the operator can input correction data manually.

[Effects of the Invention] According to the present invention, character data of the read result and grayscale image of the form data can be displayed simultaneously, so the negative phase when the operator makes corrections is significantly reduced, and the correction value of the correct character is reduced. It is also possible to refer to the preceding and following characters, making it easier to correct each word. Furthermore, by displaying form data as a grayscale image on the display, the pixel density can be reduced with almost no reduction in resolution compared to binary display, making it possible to display one line or one page of the form on the display. Become. This eliminates the disadvantage that when images are displayed in a binary pattern, the display area becomes large and it becomes difficult to display one line or one page of a form due to the display space.

[Embodiments of the invention]

Hereinafter, the present invention will be explained with reference to the drawings.

FIG. 2 shows a block diagram of a character reading device according to an embodiment of the present invention. In this figure 2, the reading form (
The character group of "power" is scanned by a rusk scan in the photoelectric conversion unit (1), photoelectrically converted, and binarized.Then, the image signal of one sentence of the obtained image signal is cut out, and the image signal of one sentence is cut out, and the reading unit ( Character recognition is performed in this reading section (2), and the character code is stored in the reading result buffer (3).On the other hand, the photoelectric conversion section (1)
The output of is also sent to the gradation circuit (8). This gradation circuit (8) extracts TIXn pixels from the two-dimensionally arranged input 1iiri image data, calculates the density value for one pixel of the corresponding gradation image, and performs data compression.

FIG. 3 specifically illustrates an example of this shading. In this example, the input image is placed on the binary image plane a3 with white=0. Black = 1, and this is shown as data arranged two-dimensionally in a matrix. Then, the sum of the data of 2 pixels of 2 sentences in the data array + 141 is calculated as 1 pixel (value of 10) on the corresponding grayscale image plane 09. Various methods can be considered, such as sum.

This is an indication. As can be seen by comparing these three figures, "7" can be most clearly distinguished in the grayscale image of FIG. 6. The density value calculated in this way is stored in the image buffer (9) in FIG. Since this density value has a lower pixel density than a binary image and is grayed out, less memory is required to store it in the image buffer 19). The output of this image buffer (9) is DA-converted by a video signal converter 0, and further converted into a video signal. This video signal and the video number output from the character code formatted in the reading result buffer (3) are combined and sent to the display (4). Refer to the lower screen 02 on which the character of the reading result is displayed on the upper screen (old) of this display (4), and then
The display data of υ is corrected by operating the keyboard (6), for example, from "?" to "3".

FIG. 7 is a flowchart showing an example of the operation of the controller (5). The controller (5) drives the photoelectric conversion unit (1), scans one line of character strings on the form, and outputs a binary signal. This binary signal is sent to the reading section (2), stored in a one-line buffer (not shown), and also supplied to the gradation circuit (8). The gradation circuit (8) is 2
The signal signal is converted into a multilevel signal using the method described above and stored in the image buffer (9).

Next, when the controller (5) drives the reading unit (2), the reading unit (2) cuts out the character string image in the one-line buffer character by character, sequentially performs recognition processing, and reads the reading results. The results are stored in the result buffer (3). Controller (5
) displays the contents of the reading result buffer (3) in sequence at the top of the display (4). On the other hand, the controller (5) converts the multi-level signal of the image buffer (9) into a grayscale video signal using the video signal converter On,
) is displayed at the bottom of the screen. That is, in the upper part of the display (4), the reading result for one line is displayed, and at the lower part, a gray scale image of the character string for one line to be read is displayed.

When the operator looks at the displayed results and finds an erroneous or eject character in the read result, he gives a correction signal to the controller (5) by pressing a specific key on the keyboard (6). This controller (5) performs a correction process,
Wait for the keyboard (6) to supply the correction location and correct character code. By looking at the grayscale image, the operator can easily determine the correct character based on its context, etc., and then presses the key corresponding to that character code.

The controller (5) corrects the reading result using correction characters given from the keyboard (6).

When the correction is completed in this way or when no correction is required, the controller (5) drives the photoelectric conversion unit (1) again to scan the next one-line character string, Similar processing is performed thereafter.

FIGS. 8 and 9 show other embodiments of the present invention. In the ANS diagram, the binary data output from the photoelectric conversion unit (1) is stored as is on the image buffer surface. Instead, there is a gradation circuit (8) on the output side to compress data and create a gradation image.

In this example, a one-line buffer (not shown) in the reading section (2)
Since the image buffer OD can be shared with the controller (5), the number of nodes can be reduced compared to the previous embodiment. FIG. 9 shows an example in which a storage device 1 is further connected to the controller (5). In this example, the documents are read one after another without operator correction when reading the documents, and the document image data that is the contents of the image buffer and the character code that is the recognition result are stored in the storage device Ogl. , the contents of the memory device l'l mallet are read out and displayed on the display (4), and the operator performs correction operations. In this case, a portable storage medium, such as an FD,
IVIT! If a movable disk or the like is used, it becomes possible for the operator to perform correction operations using another device.

In this way, in this embodiment, batch correction is possible, and system usage efficiency is greatly improved.

As detailed above, the present invention can be applied in various ways and is not limited to the above-described embodiments. For example, various display methods are possible for the output format of the display (4) shown in FIG. It is also possible to make modifications such as linking a part of the grayscale image corresponding to the target character, changing the density, or changing the color to clarify the correspondence and improve operability.

[Brief explanation of the drawing]

Fig. 1 is a hardware block diagram of a conventional character reading device, Fig. 2 is a diagram showing an embodiment of the present invention, Fig. 3 is a diagram showing an example of shading of image data, Fig. 4 is a binary image display diagram of "71", Figure 5 is a thinned-out diagram of "7" in Figure 4,
Figure 6 is a multi-level grayscale image of "7" in Figure 4.
The figure is a chart showing the flow of controller operation, and FIGS. 8 and 9 are block diagrams showing other embodiments of the invention, respectively. DESCRIPTION OF SYMBOLS 1... Photoelectric conversion part, 2... Reading part, 3... Reading result buffer, 4... Display, 5... Controller, 6... Keyboard, 7... Reading form, 8...・
- Graying circuit, 9... Image buffer, 10...
Video signal converter, 11...Reading eyelid display screen, 12
...Form image display screen, 13...Binary image screen,
14... Example of nXn pixels, 15... Grayscale image plane, 1
6... 1 pixel corresponding to nXn pixels, 17... Image buffer, 18... Storage device. Agent Ben Misaki Kensuke Chika (and 1 other person) 484

Claims (1)

[Scope of Claims] Means for scanning a group of characters to be read to obtain a binary image signal; reading means for reading the group of characters from the binary image signal obtained by this means; A gradation means for obtaining gradation image data obtained by compressing and converting a signal for a plurality of pixels constituting a value image signal into a signal for one pixel by a means for assigning predetermined gradation weighting to a signal for a single pixel; A character reading device comprising display means for displaying the result and displaying the gray scale image data of the character group obtained by the gray scale means.