JPH0432985A - Character recognition device - Google Patents

Abstract

(57) [Summary] This bulletin contains application data before electronic filing, so abstract data is not recorded.

Description

[Detailed Description of the Invention] [Object of the Invention] (Industrial Application Field) The present invention relates to a character recognition device that recognizes characters from a manuscript, and in particular, to reliably perform character recognition no matter which direction the manuscript is read. This invention relates to a character recognition device that can perform

(Prior Art) In a conventional character recognition device, as shown in FIG. 5, a reading unit 501 first reads a manuscript in which various documents are written. At this time, the reading unit 501 does not treat the document as a collection of characters, but reads it as image data consisting of a collection of dots. Then, the character recognition unit 502 cuts out an arbitrary range from the image data read by the reading unit 501, and performs pattern matching processing with standard character patterns stored in a recognition dictionary (not shown). As a result, if the character recognition unit 502 can match the character, the character is transferred to the output unit 503.
It is output to the storage unit 504 and stored in the storage unit 504. By performing these processes, characters on the document are input to the device. Note that the control section 505 controls each section. In conventional character recognition devices, the pattern matching process is performed on the assumption that the vertical position of the document has been determined. Therefore, there is a problem in that the characters on the document cannot be correctly recognized unless the document is placed in a predetermined direction and read by the reading section 501.

(Problems to be Solved by the Invention) As described above, conventional character recognition devices have a problem in that if the document is placed in the wrong direction, the characters in the document cannot be correctly recognized.

The present invention has been made to solve the above-mentioned problems, and an object of the present invention is to provide a character recognition device that can correctly recognize characters in a document regardless of the orientation of the document.

[Structure of the Invention] (Means for Solving the Problems) In order to achieve the above-mentioned object, the character recognition device of the present invention includes a reading section for reading the written content on the paper, and a predetermined method for the reading result by the reading section. a character recognition unit that performs character recognition from each of a plurality of directions, and a direction determination unit that selects a recognition result from a valid direction for the reading result by comparing the character recognition results from each direction by the character recognition unit. Equipped with

(Function) According to the character recognition device of the present invention configured as described above, the orientation determination unit determines the orientation of the original, so that even if the orientation of the original is incorrect, the characters in the original can be correctly recognized. can do. Therefore, the operator does not have to worry about the direction in which the document is placed, which reduces the burden on the operator when inputting the document.

(Example) Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 shows a block diagram of an embodiment of the present invention.

The following is an example of a case where the manuscript is written in English.

In FIG. 1, a reading unit 101 reads a document, and uses, for example, a scanner equipped with a line image sensor. Note that the type of characters that can be read is determined by the resolution of this scanner, and the resolution is determined by the type of characters on the document to be read.

The output unit 102 uses a display or the like for displaying the results of pattern matching processing, which will be described later, performed on the document read by the reading unit 101. Note that the screen of the display is divided into two parts, and one side is connected to the reading section 10.
It is also possible to display the image read by one device, and to display the corresponding characters recognized by the device on the other device.

The character recognition unit 103 recognizes the image data obtained as a result of the reading unit 101 reading the original as characters. (Here, image data is recognized from multiple directions, which will be described later.) When the reading unit 101 reads a document, it does not read character by character, but scans the document sequentially from one end. (scanning), and image data is obtained by representing each point as a binary value indicating whether there is information or not. For this reason, the higher the resolution, the more finely the document will be read, making it possible to read even more complex characters. Then, the character recognition unit 103 compares an arbitrary range of the image data (that is, cuts out each character or word) and performs a pattern matching process with a standard character pattern stored in advance by the character recognition unit 103.

This pattern matching process is performed on all of the image data, and characters corresponding to these are recognized.

The direction determining section 104 compares the character recognition results in each direction from the character recognition section 103, and determines the direction of the document based on the results. For example, the reading unit 101 reads a document, and the character recognition unit 103 performs character recognition on the image data of the document from multiple directions, and calculates the recognition rate indicating how many characters have been recognized. Compare each direction. Then, the orientation of the document is determined by setting the position where character recognition is started in the direction with the highest recognition rate as the upper left edge of the document.

The storage unit 105 stores image data read by the reading unit 101, character recognition results by the character recognition unit 103, and the like. The control section 106 controls each section and mediates data.

FIG. 2 is a flowchart showing the operation of the control section 106. When an operator sets a document (here, it is assumed to be an English document) that the operator wants the device to read into the reading unit 101, the device starts operating. Once the original is set, it will be read and
The scanning unit 101 sequentially scans the document starting from one end, and image data of the document is obtained (step 5201). This image data is then stored in the storage unit 105. The character recognition unit 103 sequentially recognizes the characters of the image data stored in the storage unit 105 while performing pattern matching processing with standard character patterns in the stored dictionary (step 5202). Note that the normal direction in which the pattern matching process is performed at this time (the recognition start position is the upper left end of the document, and this upper end is recognized as the upper side of the document) will hereinafter be referred to as the forward direction. Also, find the character recognition rate in the forward direction and set this as A (step 820
3). Next, pattern matching processing is performed on the image data in the forward and reverse directions (the recognition start position is the lower right edge of the original, and this lower edge is recognized as the upper side of the original);
Characters are recognized (step 5204). This means that the image data obtained by scanning the document by the reading unit 101 is stored in the storage unit 101.
Consider the case where the addresses are sequentially stored in descending order of addresses in 05.

The case where character recognition is performed on this image data in descending order of addresses is defined as a forward direction. In other words, the storage unit 105
Character recognition is performed in the order in which they are stored.

On the other hand, if the pattern matching process is performed in ascending order of the addresses in the storage unit 105, characters will be recognized in the opposite direction, that is, in the reverse direction. (However, the upper part of the image data is reversed in descending order and ascending order.) Then, find the character recognition rate in the reverse direction and set it as B (step S
205). (Note that if the image data is stored in the character recognition unit 103, the character recognition unit 103 can perform character recognition.) This recognition result in the forward direction, recognition rate A, and recognition result in the reverse direction. and the recognition rate B, the direction determination unit 104
send to The direction determining unit 104 compares the recognition rates of the received forward and reverse recognition results and recognition rates (step 5206). As a result, if the recognition rate A in the forward direction is higher than the recognition rate B in the reverse direction, the forward direction is the correct orientation of the document, and the recognition result in the forward direction is output to the output unit 102 as the final recognition result (step 5107). . Conversely, if the recognition rate A in the forward direction is lower than the recognition rate B in the reverse direction, this indicates that the reverse direction is the correct document orientation.

In other words, the original was read upside down (the operator set the original upside down). Then, the recognition result in the opposite direction is output to the output unit 102 as the final result (
Step 8208). By performing the above-described processing, the operator can cause the reading unit 101 to read the document without worrying about the top or bottom of the document.

Next, as shown in Figures 3 and 4, "I have ap
A case will be described in which a manuscript written in English as ``en'' is subjected to character recognition by the character recognition device of the present invention. Note that FIGS. 3 and 4 show examples of output from the output unit 102, with FIG. 3 showing the case where the document is read in the reverse direction, and FIG. 4 showing the case where the document is read in the correct direction. First, the third
As shown in Figure (a), the document is read in the reverse direction by the
If you read it and recognize the characters in the forward direction,
As shown in Figure 3(b), “-u#d # ###
If the manuscript in Figure 3 (a) is read in a literary journal in the reverse direction, it will be recognized as “y I” as shown in Figure 3 (e).
Ihave apen, is recognized. In addition"#"
is a symbol indicating that the text could not be recognized as a character.

In this way, when character recognition is performed on the document shown in FIG. 3(a), the recognition rate in units of characters is about 40% of the recognition rate A in the forward direction, as shown in FIG. 3(b). Furthermore, if standard word patterns are stored in addition to standard character patterns in the dictionary of the character recognition unit 103, the recognition rate can be calculated for each word. When word patterns are stored in the character recognition unit 103, the recognition rate for each word in FIG. 3(b) is approximately 2.
It will be 5%.

Next, as shown in FIG. 3(C), when the original of FIG. 3(a) is character-recognized in the reverse direction, the recognition rate B is 100% for both character and word units. The direction determining unit 104 compares these recognition rates A and B, and determines that the direction shown in FIG. 3(C), that is, the reverse direction, is the correct orientation of the document, and
) is output to the output unit 102 as the final recognition result and is also stored in the storage unit 105.

Furthermore, a case will be described in which characters of the original shown in FIG. 4(a) are recognized by this apparatus. Note that Fig. 4(a) is similar to Fig. 3(a).
is read in the opposite direction (set in the opposite direction). Therefore, contrary to the explanation of Fig. 3, the recognition rate when character recognition is performed in the forward direction as shown in Fig. 4 (b) is higher than that in the reverse direction as shown in Fig. 4 (C). recognition rate. As a result, it is determined that the forward direction is the correct orientation of the document, and the recognition result shown in FIG.

As described above, when the reading unit 101 reads a document, it can be read without worrying about the orientation of the document. Therefore, even if the operator sets the original in the wrong direction, the time required to replace the original and perform character recognition again can be reduced. This effect becomes greater as the number of originals whose characters are recognized by this apparatus increases.

Note that the present invention is not limited to the embodiments described above. For example, in the above embodiment, character recognition was performed on the image data of the document read by the reading unit 101 from two directions, the forward direction and the reverse direction, but the horizontal direction (direction perpendicular to the forward direction and the reverse direction), etc. It is also possible to perform character recognition from other directions and thereby determine the direction of the document.

This allows character recognition to be performed without the operator having to classify the manuscripts even when horizontally written manuscripts such as English and vertically written manuscripts such as Japanese are mixed together. Further, instead of performing character recognition in the forward and reverse directions on all of the image data of the read document to determine the recognition rate, character recognition is first performed on some image data in both the forward and reverse directions. It is also possible to determine the direction based on this recognition rate and perform character recognition for the entire image data in this direction. At this time, if it is determined that the manuscript is written in a language that only has horizontal writing, such as English, there is no need to recognize the manuscript from four directions, and the orientation of the manuscript can be recognized in at least two directions (forward and reverse). You can also decide.

Furthermore, in this embodiment, character recognition is always performed on image data from multiple directions to determine the orientation of the document, but only when the recognition rate in the forward direction is below a certain value, character recognition is performed from the reverse direction. You can also determine the orientation of the document. Furthermore, in this embodiment, a case has been described in which characters are recognized in a manuscript written in English, but by changing the standard character patterns and word patterns stored in the character recognition unit 103, it is possible to recognize characters in other languages. It can also perform character recognition. Furthermore, if character recognition is performed on the image data of the read document from multiple directions, and the recognition rate for each does not reach a predetermined value (in other words, the operator accidentally set the document upside down), Direction determination unit 104
It is also possible to give an instruction to the reading unit 101 to read the other side of the document. As a result,
The reading unit 101 can also read the other side of the document and perform the above-described character recognition on this image data.

Note that if the reading unit 101 does not have the function of reading a document from both sides, the output unit 102 may display a message to the operator to turn over the document. This is also possible by providing the reading unit 101 with a mechanism that mechanically turns over the document.

In addition, in this embodiment, character recognition of image data is performed in the forward or reverse direction by operating the address of the storage unit 105 that stores the image data, but the scanning direction of the document by the reading unit 101 itself can be changed. The orientation of the document can also be determined by obtaining image data in each scanning direction and determining the recognition rate for each image data. In this case, address operations in the storage unit 105 are unnecessary.

In short, the present invention can be implemented with various changes without departing from the spirit thereof.

C Effects of the Invention] As described above, according to the character recognition device of the present invention, the characters on the document can be correctly recognized no matter what direction the document is set. You can set it without having to worry about it, reducing the burden when inputting documents.

[Brief explanation of the drawing]

FIG. 1 is a block diagram showing the configuration of an embodiment according to the present invention, FIG. 2 is a flowchart showing the processing of the control section, and FIG.
4 are diagrams used to explain character recognition of a manuscript written in English, and FIG. 5 is a block diagram of a conventional character recognition device. 101... Reading section, 102... Output section, 103
...Character recognition section, 104...Direction discrimination section, 105 River storage section

Claims (6)

[Claims] (1) A reading section that reads the written content on the paper, a character recognition section that performs character recognition from multiple predetermined directions on the reading results from this reading section, and a character recognition section that recognizes the character recognition results in each direction by this character recognition section. A character recognition device comprising: a direction determination unit that selects a recognition result from a valid direction with respect to the reading result by comparison. (2) The character recognition device according to claim 1, wherein the direction determining section selects a character recognition result having the highest recognition rate among the character recognition results in each direction. (3) The direction determining section requests the reading section to read the back side of the paper when the character recognition result in each direction by the character recognition section does not reach a predetermined value. The character recognition device described. (4) The character recognition unit performs character recognition in the forward direction according to the reading order by the reading unit, and character recognition in the reverse direction of the reading order with the upper side of the page reversed from that in the forward direction. The character recognition device according to claim 1. (5) The character recognition unit performs character recognition in a direction orthogonal to the forward and reverse directions when the recognition rate in the forward and reverse directions does not reach a predetermined value. Character recognition device. (6) The character recognition unit performs character recognition on a portion of the reading results from multiple directions, and the direction determining unit determines a valid direction in which the reading results should be recognized according to these character recognition results. 2. The character recognition device according to claim 1, wherein the character recognition device comprises: