JPH0420506B2 - - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of Industrial Application] The present invention relates to a character input device that converts a character image in a digital image into a character code and inputs the converted character code into a computer.

[Conventional technology]

An image file system is known in which a document consisting of text and drawings is digitally converted using an image sensor, the resulting digital image is input into a computer, and is stored on a hard disk. The character image is a computer
Storing it as a digital image in RAM or hard disk would occupy a large amount of memory space, so it is desirable to store it as a character code. If characters are converted into character codes, they can be stored in a memory area many times smaller. Also, in image data processing, there will be no interference with calculations.

For this reason, a digitizer is a conventional device for separating figures and characters written on a document in a digital image. This method involves manually entering the required position in the document using a cursor.
Specify the range in which characters exist. Further, the content of the characters is separately specified and inputted by the user using the keyboard.

FIG. 8 is an explanatory diagram of character input using a conventional character input device consisting of a digitizer and a keyboard.
In the figure, 11 is a board on which documents are placed, 12 is a document including a target document and a drawing, 13 is a cursor for inputting a position, 14 is a switch, and 15 is a keyboard.

Next, the operation will be explained.

First, the target document 12 is pasted on the board 11 and the point to be input is specified using the cross mark of the cursor 13. Then, the coordinates are input using the switch 14.

In this case, two diagonal points of the rectangle where the character exists are input.

In this way, the digital image of the memory area occupied by the character is removed, the memory area is left blank, and the data of the character's code is set to be stored.

Next, the user inputs what the target character is into the computer using the keyboard 15 as a character code.

[Problem that the invention seeks to solve]

Conventionally, when inputting characters existing in a document into a computer, the code information for the characters is input after manually creating a blank space in the memory, as in the above-mentioned device, which takes time. In addition, there were other problems such as input errors.

This invention was made to solve the above-mentioned problems, and provides a character input device that can convert character information in a document into a character code and input it into a computer with simple operations and without input errors. The purpose is to

[Means for solving problems]

The present invention comprises an image input means 1 that optically reads a document containing characters and figures to generate digital image data, an image storage means 2 that stores this digital image data in a memory consisting of a large number of pixels, and Character separating means 3 is given stored digital image data and separates it into graphic image data and character image data; Character reading means 4 is given this character image data and compares it with dictionary data to recognize a character code; A character array means 5 for determining the direction of a character string from character string image data consisting of a plurality of character image data and rotating and arranging the character string image data so that the direction of the character string is horizontal; and the arranged character string. Display means 6 for displaying image data and character data of a character code in correspondence, and interaction correction means 7 for inputting a character code so that the character data matches the displayed character string image data.
It consists of.

[Effect]

Character image data is automatically separated from a digital image by a character separating means 3. Further, the character reading means 4 automatically converts the character code from the character image data. Further, the character array means 5 determines the direction of the character string from the character string image data consisting of a plurality of character image data, and rotates and aligns the character string image data so that the direction of the character string becomes horizontal. If there is an error in the character code conversion, it is manually corrected on the display means 6 by the dialogue correction means 7.
Character separation is automated and processed at high speed, and character code conversion is guaranteed to be fast and accurate.

〔Example〕

Hereinafter, one embodiment of the present invention will be described based on the drawings.

FIG. 1 is a block diagram showing the overall structure of a character input device according to the present invention for inputting document data into a computer. In the figure, 1 is an image input means such as an image sensor that optically reads documents including drawings and documents and captures binary image data, and 2 is a pixel memory that stores digital image data input by this image input means 1. 3 is a character separation means for processing the document image data stored in the image storage means 2 to obtain image data of a character portion; 4 is image data of this character portion (hereinafter referred to as character image data) A character reading means performs character recognition processing on the character to obtain a code (hereinafter referred to as a character code) representing what character it represents; 5 is a character for organizing the arrangement of character image data obtained by the character separation means 3; Arranging means 6 is a display means for comparing and displaying the character code obtained by the character reading means 4 and the character image data arranged by the character arranging means 5; 7 is a display means for displaying the character code displayed by the display means 6; It is a dialogue modification means having a keyboard etc. to be modified.

Next, the operation of the character input device will be explained with reference to FIGS. 2 to 7.

First, the image input means 1 generates document image data. For example, the character “23” as shown in Figure 2
Binary image data consisting of and a figure is obtained and stored in the image storage means 2. Next, character separation means 3
In order to extract character image data from the image data stored in the pixel memory of the image storage means 2,
First, as shown in FIG. 3, the same symbol ("1" and "2") is assigned to each set (island image) of connected detection pixels (hereinafter simply referred to as black dots).

Next, as shown in FIG. 4, the character separating means 3 selects the horizontal and vertical sizes A,
Find B. The number C of black points for the rectangular area A×B is calculated, and the ratio of these A, B, and C is C/(A×
B) (where C is the total number of black dots) is used as a parameter to determine whether or not it is a character. If it is determined to be a character, the above set of black dots "1" is output as character image data.

In the same way, the characters of the set of black dots "2" are judged and outputted. After that, another set of black dots is examined, all character candidates in the document are judged, and separated and output. In this way, each character image data is removed from the rectangular area (A×B, etc.) of the pixel memory occupied by each character image data, and a blank area having a width of A×B, etc. is generated. Each character image data is then applied to each buffer memory of the character reading means 4 and the character arrangement means 5 and stored therein. Character dictionary data is stored in advance in the character reading means 4, and the size of the character image data is reduced or enlarged to match the size of the dictionary data on the buffer memory.

The character reading means 4 then identifies what character the character image data obtained by the character separation means 3 is.

As shown in Fig. 5, the identification method is to correlate the character image data of the matched sizes with the character image data of a dictionary of a predetermined size prepared in advance (hereinafter referred to as dictionary data), and calculate the evaluation value V. The code in the dictionary with the maximum value is recognized as a character code and stored in the built-in register.

The evaluation value V is determined by looking at character image data (hereinafter referred to as normalized data) and dictionary data whose sizes are matched, and counting the pixels where both are black points (referred to as a). Also,
Count the number of black dots in the normalized data (let it be b) and the number of black dots in the dictionary data (let it be k) k/(a×b)
is the evaluation value V.

Further, in the character arrangement means 5, the arrangement of character image data is rotated and rearranged. As shown in Figure 6, this is done by treating adjacent character image data on the buffer memory as one character string, and starting from the center of the character string.
Draw a straight line every 10°. Count the points passing through the sunspots for each straight line, and calculate the distance between the two points that are farthest apart, the distance between the farthest sunspots. The distance between the farthest sunspots for each straight line is compared, and the straight line with the greatest distance is set as the direction of the character string. Next, the character image data is rotated so that the direction of the character string is parallel to the pixel direction on the buffer memory, and the character arrangement is rearranged. This rotated character string is stored as character image data in a built-in register. The contents of the register of the character reading means and the register of the character string arranging means 5 are given to the display means 6.

In the display means 6, as shown in FIG. 7, for a plurality of character strings at once, the character image data of the organized character strings and the character data corresponding to the character code obtained by the character reading means 4 for each character are arranged side by side. indicate. Since some of the characters in the document are out of shape and the character recognition by the character reading means 4 is not all correct, the original character image data is visually checked by the operator in an interactive manner.

Using the dialogue correction means 7, compare the character image data and character data of the character string displayed by the display means 6, and if there is an error, change the character "C" in the erroneous character string. Pick on the tablet of the dialog correction means 7 and input the correct character code "C" from the keyboard. Most of the character image data has been correctly converted into character codes, and only a few errors need to be corrected.

In addition, if the character reading means 4 is not made to function,
Only the character image data of the organized character string is displayed on the display means 6. Looking at the character image data of the displayed character string, the user reads the corresponding characters from the keyboard of the dialogue modification means 7 and inputs the character code. Since characters and figures are separated, only character code conversion is done manually.

These character codes become the input of character information to the computer. In this way, character images in digital image data can be easily separated from drawing images, and character image data can be converted into character codes at the same time.

〔Effect of the invention〕

As explained above, according to this invention,
an image input means for optically reading a document containing characters and figures to generate digital image data; an image storage means for storing this digital image data in a memory consisting of a large number of pixels; character separating means for separating given graphic image data into character image data; character reading means for receiving this character image data and comparing it with dictionary data to recognize a character code; and a character string consisting of a plurality of character image data. character arrangement means for determining the direction of the character string from the image data and rotating and arranging the character string image data so that the direction of the character string is horizontal; and the arranged character string image data and the character data of the character code. and an interactive correction means for inputting a character code to match the displayed character string image data with the displayed character string image data. It is possible to check and correct character recognition results and input character information into a computer at high speed with less effort. Furthermore, even if the character string is written diagonally, the character string image data is rotated and displayed so that the direction of the character string is horizontal, making it easier to read the characters in the character string image data. It is possible to quickly check and correct character recognition results and input character information into a computer.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of a character input device according to the present invention, FIG. 2 is an explanatory diagram of digital image data, and FIG. 3 is an explanation of preprocessing for separating characters in the character separating means. Figure, 4th
The figure is an explanatory diagram of reference parameters for determining characters in the character separation means, Figure 5 A and B are explanatory diagrams for identifying characters in the character reading means, and Figure 6 is an explanation of organizing the character arrangement in the character arrangement means. 7A and 7B are explanatory diagrams showing display examples on the display means and dialogue modification by the dialogue modification means. 8th
The figure is an external view of a conventional digitizer for inputting characters. 1... Image input means, 2... Image storage means, 3
...Character separation means, 4...Character reading means, 5...
Character arrangement means, 6...Display means, 7...Dialogue modification means.

Claims (1)

[Claims] 1. An image input means for optically reading a document containing characters and figures to generate digital image data;
image storage means for storing this digital image data in a memory consisting of a large number of pixels; character separation means for receiving this stored digital image data and separating it into figure image data and character image data; A character reading means that recognizes a character code by comparing it with given dictionary data, and a character string image that calculates the direction of a character string from character string image data consisting of a plurality of character image data so that the direction of the character string is horizontal. a character arrangement means for rotating and arranging data; a display means for displaying the arranged character image data and the character data of the character code in correspondence; and a display means for matching the character data with the displayed character string image data. A character input device comprising a dialogue correction means for inputting a character code.