JP2009053826A - Document processor and document processing program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a document processor and a document processing program for discriminating handwritten characters and printed characters without using any character image with high resolution. <P>SOLUTION: When a character image acquisition part 22 acquires a character image as the object of determination, a line segment extraction part 24 executes filamentizing processing to the acquired character image, and extracts line segments from the character image. In this case, the line segment extraction part 24 may extract an end point and an intersection point from the character image, and decompose the character for every line segment on the basis of the edge point and intersection point to extract the line segments. A curvature variation arithmetic part 26 calculates curvature variation for every line segment extracted by the line segment extraction part 24. A character discrimination part 28 totals the curvature variation calculated by the curvature variation arithmetic part 26, and generates a histogram, and determanates whether or not a measured number in the low pass section of the histogram exceeds a preset threshold. When the measured number exceeds the threshold, the character discrimination part 28 discriminates that the object character is a handwritten character, and when the number of measurement does not exceed the threshold, discriminates that the object character is a printed character. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to a document processing apparatus and a document processing program.

Conventionally, techniques for discriminating handwritten / printed characters from character strings in a document have been proposed. For example, in the following Patent Document 1, attention is paid to character outline information, and whether a character outline is linear is determined by distinguishing between handwritten characters and printed characters. In this technique, high resolution is required to clearly extract the contour line information.
JP 58-37776 A

  An object of the present invention is to provide a document processing apparatus and a document processing program for discriminating handwritten characters and printed characters without using a high-resolution character image.

  In order to achieve the above object, the document processing apparatus according to claim 1 is characterized in that a line segment extracting unit that extracts a line segment from a character image and a curvature change amount calculating unit that calculates a change amount of the curvature of the line segment. And character discriminating means for discriminating between handwritten characters and printed characters based on the amount of change in curvature.

  The invention according to claim 2 is characterized in that, in the invention according to claim 1, the line segment extraction means performs line segment extraction after thinning a line segment constituting a character.

  The invention according to claim 3 is the invention according to claim 2, wherein the line segment extracting means extracts end points and intersections of line segments constituting the character, and decomposes the character into line segments at these points. It is characterized by extracting a line segment.

  The invention according to claim 4 is the invention according to any one of claims 1 to 3, wherein the character discriminating means discriminates a handwritten character and a printed character from the curvature change amount histogram. And

  The invention according to claim 5 is the invention according to claim 4, wherein the character discriminating means discriminates a handwritten character and a printed character from the peak shape of the histogram.

  The document processing program according to claim 6 is a computer program comprising: a line segment extracting means for extracting a line segment from a character image; a curvature change amount calculating means for calculating a change amount of the curvature of the line segment; and a change amount of the curvature. It is made to function as a character discrimination means which discriminate | determines a handwritten character and a printed character based on this.

  According to the first aspect of the present invention, it is possible to discriminate between a handwritten character and a printed character without using a high-resolution character image as compared with the case of using character outline information.

  According to the invention of Claim 2 and Claim 3, compared with the case where it does not have this structure, a line segment can be extracted from a character image easily.

  According to the invention of Claim 4 and Claim 5, compared with the case where this structure is not provided, a handwritten character and a printed character can be discriminate | determined easily.

  According to the sixth aspect of the present invention, it is possible to provide a document processing program capable of discriminating between handwritten characters and printed characters without using a high-resolution character image as compared with the case where this configuration is not provided.

  Hereinafter, the best mode for carrying out the present invention (hereinafter referred to as an embodiment) will be described with reference to the drawings.

  FIG. 1 shows a hardware configuration of an embodiment of a document processing apparatus according to the present invention. In FIG. 1, the document processing apparatus includes a central processing unit (for example, a CPU can be used) 10, a random access memory (RAM) 12, an image reading device 14, a display device 16, an input device 18, and a hard disk device (HDD) 20. It is comprised including. These components are connected to each other by a bus 21.

  The CPU 10 controls the operation of each unit, which will be described later, based on a control program stored in the RAM 12 or the hard disk device 20. The RAM 12 mainly functions as a work area for the CPU 10.

  The image reading device 14 is configured by a scanner or the like and reads a character image.

  The display device 16 is configured by a liquid crystal display or the like, and displays a character image to be determined.

  The input device 18 includes a keyboard, a pointing device, and the like, and is used by a user to input operation instructions and the like.

  The hard disk device 20 is a large-capacity storage device that can be read by a computer, and can store various data necessary for processing to be described later.

  FIG. 2 shows a functional block diagram of an embodiment of a document processing apparatus according to the present invention. In FIG. 2, the document processing apparatus includes a character image acquisition unit 22, a line segment extraction unit 24, a curvature change amount calculation unit 26, and a character determination unit 28.

  The character image acquisition unit 22 includes, for example, the image reading device 14 shown in FIG. 1 and a program for controlling the image reading device 14 by the CPU 10, and acquires image data of a character to be determined. Note that the character image acquisition unit 22 may be configured to acquire electronic data of a character image via an appropriate communication unit or the like instead of the image reading device 14.

  The line segment extraction unit 24 includes, for example, the CPU 10 and a program that controls the processing operation of the CPU 10, and extracts a line segment from the character image acquired by the character image acquisition unit 22. The line segment can be extracted by performing a thinning process for narrowing the width of the line segment constituting the character and then tracking the line segment constituting the character. Further, for example, line segments can be extracted by extracting end points and intersections of line segments constituting the character, and decomposing the character into line segments at these points. The thinning process and the line segment extraction method will be described later.

  The curvature change amount calculation unit 26 is configured to include, for example, the CPU 10 and a program for controlling the processing operation of the CPU 10, and calculates the change amount (curvature change amount) of the curvature of the line segment extracted by the line segment extraction unit 24. The amount of curvature change can be obtained from, for example, the slope of a straight line drawn between pixels constituting the line segment. A method of calculating the curvature change amount will be described later.

  The character determination unit 28 includes, for example, the CPU 10 and a program for controlling the processing operation of the CPU 10, and determines a handwritten character and a printed character based on the curvature change amount. The discrimination between the handwritten character and the printed character can be performed by taking, for example, a histogram of the amount of curvature change. This determination method will be described later.

  3A and 3B are explanatory diagrams of the thinning process performed by the line segment extraction unit 24. FIG. FIG. 3A shows a character image (important) before the thinning process, and FIG. 3B shows a character image after the thinning process. The fine line process is a process of reducing the number of pixels in the width direction of the line segment constituting the character and converting it into a line image having a width of 1 pixel.

  FIG. 4 is an explanatory diagram illustrating an example of a line segment extraction method performed by the line segment extraction unit 24. In FIG. 4, explanation will be given by taking “a” of hiragana as an example. In FIG. 4, an end point indicated by a white circle (◯) and an intersection point indicated by a triangle (Δ) and where the line segments intersect each other exist on the line segment constituting “A”. The line segment extraction unit 24 decomposes each line segment by the end points and the intersections, and extracts a line segment sandwiched between them. At that time, for example, a line segment that is short between the end point a and the intersection point α, a line segment that is sandwiched between the intersection point α and the intersection point β, and a line segment that is sandwiched between the intersection point α and the intersection point γ are Since the determination process in the character determination unit 28 may be inaccurate, it may be excluded from the determination process. Alternatively, only the end point may be used, and a line segment sandwiched between the end points may be used for the determination process. The line segment extraction method is not limited to the example shown in FIG. Any method can be applied to the present invention as long as it can trace a line segment constituting a character.

  FIG. 5 shows an example of a method for calculating the curvature change amount performed by the curvature change amount calculation unit 26. In FIG. 5, a line segment composed of nine pixels will be described as an example. In FIG. 5, the curvature change amount calculation unit 26 sets a line connecting pixels for every adjacent pixel or by skipping several pixels, and obtains a curvature change amount from a change in inclination of the line.

  For example, when a line segment is set for each adjacent pixel, a line connecting the first pixel and the second pixel shown in FIG. 5, a line connecting the second pixel and the third pixel, and so on. A line connecting adjacent pixels is set, and a change in the inclination is obtained as a curvature change amount of the line segment. When a line connecting pixels by skipping several pixels is set, a line connecting the first pixel and the seventh pixel shown in FIG. 5 is a line connecting the second pixel and the eighth pixel. A line connecting pixels that have been skipped several pixels (five pixels skipped in the example of FIG. 5) is set, and the change in the inclination is determined as the curvature change amount of the line segment. Note that the number of pixels to be skipped is not limited.

  6A and 6B show examples of histograms used by the character discrimination unit 28 for discrimination processing between handwritten characters and printed characters. 6A and 6B, the horizontal axis represents the calculated curvature change amount, and the vertical axis represents the number of measurements (frequency) for each value of the curvature change amount. FIG. 6A is an example of a histogram of handwritten characters, and FIG. 6B is an example of a histogram of printed characters.

  In general, when the amount of curvature change of a line segment constituting a character is made into a histogram, the number of measurements of a small value (low range) of the amount of curvature change is larger for handwritten characters than for printed characters. This is because the curvature fluctuation due to handwriting appears in the low range of the curvature change amount. By comparing such differences in histograms, it is possible to distinguish between handwritten characters and printed characters. For example, in the histogram of handwritten characters shown in FIG. 6A, the number of measurements in the low range of the curvature change amount is larger than the histogram of type characters shown in FIG. 6B. Therefore, an appropriate threshold value is set, and when the number of measurements in the low range of the curvature change amount exceeds the threshold value, it can be determined as a handwritten character.

  In addition, when writing characters by hand, it is difficult to draw a pure straight line or a curve with a certain curvature. Therefore, handwritten characters are more likely to be blurred, that is, have smaller meanders than printed characters. For this reason, in the histogram of handwritten characters, a peak having a certain width is formed in the low band as shown in the circled part of FIG. On the other hand, a peak with a small width appears in the histogram of type characters. Therefore, it is possible to discriminate between handwritten characters and printed characters by looking at the peak shape.

  As described above, the character discriminating unit 28 receives the curvature change amount calculated by the curvature change amount calculating unit 26, obtains the number of measurements for each value, and displays the histograms shown in FIGS. 6 (a) and 6 (b). By generating, it distinguishes between handwritten characters and printed characters.

  FIG. 7 shows a flow of an operation example of the document processing apparatus according to the present invention. In FIG. 7, the character image acquisition unit 22 acquires a character image as a discrimination target (S1). The character image is acquired by, for example, a process in which a user sets the image reading device 14 and inputs a reading instruction from the input device 18.

  The line segment extraction unit 24 performs thinning processing on the character image acquired by the character image acquisition unit 22 (S2), and extracts a line segment from the character image (S3). Here, the line segment extraction unit 24 extracts end points and intersections from the thinned character image, and extracts line segments from the character image by decomposing the characters into line segments based on the end points and intersections. Good.

  The curvature change amount calculation unit 26 calculates a curvature change amount for each line segment extracted by the line segment extraction unit 24 (S4).

  The character discriminating unit 28 totals the curvature change amounts calculated by the curvature change amount calculating unit 26 and generates a histogram (S5). Next, the character determination unit 28 determines whether or not the number of measurements in the low frequency portion of the histogram described with reference to FIGS. 6A and 6B exceeds a preset threshold value (S6).

  The character discriminating unit 28 discriminates that the target character is a handwritten character when the measured number exceeds the threshold (S7), and discriminates it as a printed character when the measured number does not exceed the threshold (S8).

  Although several embodiments of the present invention have been introduced above, the present invention is not limited to the above embodiments.

It is a figure which shows the hardware constitutions of one Embodiment of the document processing apparatus concerning this invention. It is a functional block diagram of one Embodiment of the document processing apparatus concerning this invention. It is explanatory drawing of the thinning process which a line segment extraction part performs. It is explanatory drawing of an example of the line segment extraction method which a line segment extraction part performs. It is a figure which shows the example of the calculation method of the curvature change amount which a curvature change amount calculation part performs. It is a figure which shows the example of the histogram which a character discrimination | determination part uses for the discrimination | determination process of a handwritten character and a printed character. It is a flowchart of the operation example of the document processing apparatus concerning this invention.

Explanation of symbols

  10 CPU, 12 RAM, 14 image reading device, 16 display device, 18 input device, 20 hard disk device, 21 bus, 22 character image acquisition unit, 24 line segment extraction unit, 26 curvature change amount calculation unit, 28 character discrimination unit.

Claims (6)

A line segment extracting means for extracting a line segment from a character image;
Curvature change amount calculating means for calculating a change amount of the curvature of the line segment;
Character discriminating means for discriminating between handwritten characters and printed characters based on the amount of change in curvature;
A document processing apparatus comprising:   2. The document processing apparatus according to claim 1, wherein the line segment extraction unit performs line segment extraction after thinning a line segment constituting a character.   3. The document processing apparatus according to claim 2, wherein the line segment extracting unit extracts end points and intersections of line segments constituting the character, and extracts the line segment by decomposing the character into line segments at these points. A document processing apparatus characterized by the above.   4. The document processing apparatus according to claim 1, wherein the character discriminating unit discriminates between a handwritten character and a printed character from the curvature change amount histogram. 5.   5. The document processing apparatus according to claim 4, wherein the character discriminating unit discriminates handwritten characters and printed characters from the peak shape of the histogram. Computer
A line segment extracting means for extracting a line segment from a character image;
Curvature change amount calculation means for calculating a change amount of the curvature of the line segment,
Character discriminating means for discriminating handwritten characters and printed characters based on the amount of change in curvature,
A document processing program characterized by functioning as

Images