JPS58105385A - Character reading and recognizing device - Google Patents

Abstract

PURPOSE:To recognize the sentences belonging to the same character row, by deciding the same row if the overlap exceeds a certain value between the projection component of an independent pattern in a specific direction on an original and the projection component of an adjacent independent pattern in the specific direction. CONSTITUTION:When the characters are read and recognized to decide the sentence row to which the characters belong, an independent pattern extracting circuit 50 searches an independent patterns S3 while moving a segmenting window 50 which is larger enough than characters 51-54. A projection component inspecting circuit 43 extracts out of the signal S3 the horizontal and vertical projection components with the longitudinal writing and the lateral writing respectively. Then the circuit 43 inspects the overlap between the projection component of the immediately preceding (adjacent) characters stored in a memory and the projection component of the present characters. Then the same row is decided when the overlap is larger than the prescribed value. While a different row is decided when the overlap is smaller than the prescribed value. Then prescribed codes are given to both the same row and the different row respectively.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field of the Invention This invention relates to a character reading/identifying device (Kid), and particularly to a device for determining the text nature to which the character belongs.

In conventional optical character reading devices, lines are shown on a sheet of paper 10, as shown in FIG. 1, in order to facilitate the reading of text. - 11 was provided to restrict the range in which the 91 lines of text 12 were included from this mark 11tC.

However, according to this method, it is necessary to make marks on the paper, and it is complicated to fill in the information according to the marks. fail. In other words, as shown in Figure 2, even if the original is printed with characters aligned horizontally, if the original is set diagonally and tilted during scanning, the image data will be skewed, and if the line spacing is narrow, Horizontal projected components 1 &+ 2 &s 3 &+ as shown in the WJS diagram.
. , 101a, 102m+103m, . . . , the break cannot be detected, and therefore it is impossible to determine whether or not they are on the same line.

Purpose of the Invention The present invention was made based on the above-mentioned circumstances, and an object of the present invention is to provide a character reading recognition device capable of extracting lines even from free format text images.

To achieve this object, the present invention includes: a first device for extracting an independent pattern from a document; a second device for detecting a specific direction projection component of the independent pattern obtained by the first device; The overlap of the specific direction projection components of adjacent independent patterns is verified, and if the overlap is above a certain value, it is judged as the same line, and if it is below a certain value, it is judged as separate lines, and a predetermined code is assigned to each of them. 3.

Examples of the invention Hereinafter, the present invention will be described in detail with reference to the accompanying drawings.

FIG. 4 is a system diagram showing an embodiment of this invention, and character 1
! Weir device Tochi, isolated point removal circuit 41. , a connected pattern extraction circuit company, a projection component verification circuit 43, a row extraction circuit 44, a post-processing circuit 6, and a recognition device Tochi.

The character reading device is, for example, one of various conventionally known optical reading devices, and converts optical characters into image data S1 in units of pixels.

The isolated point removal circuit 41 is for removing noise and the like unrelated to the text, and various circuits are known. The output S2 is a picture signal from which noise has been removed.

As shown in FIG. 5, the independent pattern extraction circuit 42 searches for each independent noreen while moving through a group of cutting chambers that are sufficiently large than the characters 51 to 54. The signal 81 constitutes the pattern.

Among the independent patterns shown in FIG. 5, the letter T (51).

It is assumed that m (52) has already been extracted and given a processed code, character c (53) is the currently detected independent pattern, and character H (34) is the pattern that adheres to nitrogen 50.

The projected component verification circuit 6 extracts the horizontal projected component in the case of a vertically written document, and extracts the horizontally projected component in the case of a horizontally written document, based on the signal 88 representing the independent pattern output from the independent pattern extraction circuit. Extract the projected components.

This circuit also includes a memory, a comparator, and a code setter (none of which are shown), and verifies the overlap between the projected component of the immediately previous character stored in the memory and the projected component of the current character. Then, this overlapping value is compared with a predetermined threshold value of a comparator. In this way, if the overlap is greater than the threshold value, it is determined that they are the same line, and if the overlap is smaller than the threshold value, it is determined that they are different lines, and a predetermined code is assigned to each line.

The row extraction circuit sequentially reads signals 84 constituting an independent pattern having codes belonging to the same row among the codes given by the projection component verification circuit 43.

Based on the signal S4, the post-processing circuit 6 determines whether it belongs to the same row or not.
A predetermined sentence is displayed on the recognition device 46 using a signal S6 that forms such independent patterns. The recognition device includes various known display devices, printers, and the like.

With the above configuration, image information @81 read by the character reading device piece is subjected to unnecessary signal removal by the isolated point removal circuit 41, and independent pattern extraction circuit 42 extracts the independent pattern 8. is extracted. This pattern S
, the projected component of the pattern is extracted by the projected component testing circuit 43 and the projected components of the instantaneously touching independent patterns are tested for mutual overlap, it is determined whether they belong to the same row or not, and a predetermined code is assigned. The 0-line extraction circuit and the post-processing circuit perform classification reading of the code, and identify sentences belonging to the same character line. ! ! output via the recognition device 46.

Effects of the Invention By configuring as described above, this invention provides a character reading system that can accurately read and open the text even when the text is biased, especially in the case of a free format. identification device can be provided.

[Brief explanation of the drawing]

1 to 3 are diagrams explaining the conventional line reading method, FIG. 4 is a system diagram showing an embodiment of the present invention, and FIG. 5 shows the operation of the projected component test circuit 43 in FIG. 4. It is a figure for explaining. 40...Character reading device, 41...Isolated point removal circuit,
42... Independent pattern extraction circuit, 43... Projective component testing circuit. Necessary...Line extraction circuit, 45...Post-processing circuit, Weak...
Weaving device 1, applicant's representative Kiyoshi Inomata

Claims (1)

[Claims] a first device that extracts an independent pattern from a document; a second device that detects a specific direction projected component of the independent pattern obtained by the first device; and an overlap of the specific direction projected components of adjacent independent patterns. A character reading device comprising: a third device which performs a gate test, determines that the lines are the same line if the overlap is above a certain value, and judges that they are separate lines if the overlap is below a certain value, and assigns a predetermined code to each line. Shi recognition device.