JPH0535914A - Picture inclination detection method - Google Patents

Abstract

PURPOSE:To reduce the processing quantity by determining the inclination of an picture from a character string defined area. CONSTITUTION:A document is fetched as a binarized picture by a picture input part 101 such as a scanner, etc. For the extraction of a character string area from an input picture, the input picture is reduced in a picture reduction part 102. A block generation/classification part 103 extracts blocks by performing a gradation processing in the character string direction of the reduced picture and classifies into diagrams, photographs, character string defined areas and character string candidate area from the sizes of the blocks. The block generation/classification part 103 selects areas where the lateral lengths are more than three times as the vertical lengths in the areas extracted as character string defined areas. An inclination detection part 104 calculates a straight line whose deviation from a center point of each circumscribed rectangular is the smallest in the selected areas and the inclination of the calculated straight line becomes the inclination of the character string.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image inclination detecting method for detecting the inclination of a character string area from a document image.

[0002]

2. Description of the Related Art Conventionally, as a method of detecting the inclination of an image, a method of detecting the inclination of an image by performing a two-dimensional Fourier transform process on a document image (The 22nd National Convention of Information Processing Society of Japan, pp. 677-678). ), Or dividing the image into a plurality of strip-shaped regions, obtaining the peripheral distribution in each region, and detecting the inclination of the image from the phase shift between the peripheral distributions (The Institute of Electronics, Communication Engineers, 1983 V
ol. J66-D No. 1 pp. 111-118).

[0003]

However, each of the above-mentioned methods has a drawback in that the amount of data to be processed becomes large because the inclination of the image is obtained by processing the entire image. An object of the present invention is to provide an image inclination detection method that reduces the amount of processing by obtaining the image inclination from a character string confirmation area.

[0004]

In order to achieve the above object, the present invention comprises a step of reducing an input image in an image inclination detecting method for detecting the inclination of a character string region from a binarized document image. A step of generating a block by performing a blurring process on the reduced image and classifying a character string area from the generated block; and an inclination of the image from the position of a circumscribed rectangle of a predetermined area in the character string area. Is detected.

[0005]

The document is captured as a binary image by the image input unit such as a scanner, and the input image is reduced by the image reduction unit in order to extract the character string region from the input image. The block generation / classification unit extracts a block by performing blurring processing in the character string direction of the reduced image, and determines a figure, a photograph, a character string confirmation region from the size of the block,
The area is classified into character string candidate areas, and the block generation / classification unit selects an area having a horizontal length of three times or more the vertical length from the areas extracted as the character string finalized areas. The inclination detection unit calculates a straight line with the smallest deviation from the center point of each circumscribing rectangle in the selected area by using the least squares approximation method,
The calculated slope of the straight line becomes the slope of the character string.

[0006]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be specifically described below with reference to the drawings. 1 is a block diagram of an embodiment of the present invention, in which 101 is an image input unit for reading and inputting a document image such as a scanner, 102 is an image reducing unit for reducing the input image, and 103 is Extract the block from the reduced image, the extracted block photo, character string confirmation area,
Block generation and classification unit for classifying into character string candidate areas, 10
Reference numeral 4 is an image inclination detection unit, 105 is a data storage unit that stores data such as an input image, a reduced image, and block data, and 106 is a control unit that controls each unit.

FIG. 2 shows a processing flowchart of the present invention. In step 201, the document is captured as a binary image by the image input unit 101 such as a scanner. Next, in order to extract the character string region from the input image, the image reducing unit 102 reduces the input image. This reduction of the image is to reduce the image vertically and horizontally so that the black pixels of the upper, lower, left and right character strings do not come into contact with each other (step 202). The block generation / classification unit 103 extracts blocks by performing blurring processing in the character string direction of the reduced image, and classifies the blocks into figures, photographs, character string confirmed areas, and character string candidate areas.

Here, the blurring process is a process in which the connection of a black run and a short white run separated by a long white run is made into one run for the run length data of the image. Blocks are those in which the connected runs are connected vertically.

The block generation / classification unit 103 further selects a region whose horizontal length is three times or more the vertical length from the regions extracted as the character string finalized region (step 20).
4). By selecting an area containing multiple characters in this way, many black pixel connected components can be calculated,
Therefore, the accuracy of tilt detection can be improved.

FIG. 3 shows a character string definite area 301 whose horizontal length is three times or more the vertical length, and 302 to 308 are circumscribed rectangles of black pixel connected components. The black pixels are not the pixels on the reduced image but the pixels of the input original image.

The tilt detecting section 104 determines the center point of each circumscribed rectangle (the center point of the circumscribed rectangle 302 is 310).
The straight line with the smallest deviation from is calculated using the least-squares approximation method, and the slope of the calculated straight line is the slope of the character string to be obtained. When the straight line is calculated, each center point is weighted in proportion to the area of the circumscribed rectangle.

That is, when y (x) = a + bx, a and b are determined so as to minimize J in J = Σmi {y (xi) -fi} * 2. Here, each center point is represented by coordinates (xi, fi), and the area of each circumscribed rectangle is represented by mi. Also, * 2 represents the square. 309 of FIG.
Is a straight line calculated in this way. Then, the above processing is performed for some character string confirmation areas, and the inclination detection unit 104 obtains the inclination of the image by taking the average value of the inclinations (step 205).

<Embodiment 2> In a character string centered on the alphabet such as an English document, characters extending upward such as "d" and "l", characters extending downward such as "g" and "y", and Has a capital letter, and the error becomes large when the inclination of the image is obtained from the arrangement of the center points of the circumscribing rectangle as in the above embodiment. Therefore, the second embodiment relates to a method for correctly detecting the inclination of an image even in an English character string in which the size and position of the circumscribing rectangle vary.

FIG. 4 is a diagram showing a specific example of an English character string in which the size and position of the circumscribing rectangle vary. The inclination detection unit 104 of this embodiment creates a histogram of the height of such a circumscribed rectangle (FIG. 5), and sets the circumscribed rectangle with the highest frequency as the reference size. Then, with respect to the circumscribed rectangle that is larger (that is, higher) than the reference value by the threshold value or more, the upper portion is cut. That is, in FIG. 6, reference numeral 601 denotes the circumscribed rectangle of the black pixel connected component of the original image, and when the reference value of the height of the circumscribed rectangle obtained from the histogram of FIG. 5 is 603, the height of the circumscribed rectangle 601 is Is converted into a reference value and becomes a circumscribed rectangle 602.

In the example of FIG. 4, circumscribed rectangles 402, 403,
If 405 is a circumscribed rectangle larger than the reference value, its height is converted as indicated by 602 in FIG. Then, with respect to the circumscribed rectangle after conversion, the straight line with the smallest deviation from the center point of each circumscribed rectangle is calculated using the least-squares approximation method in the same manner as in the above-described embodiment, and thus the inclination of the character string is calculated. Ask for.

As another embodiment, the circumscribed rectangles 404 and 411, which are much smaller than the reference value of the height of the circumscribed rectangle obtained by the histogram of FIG. 5, are often periods, commas, etc. Since there is a slight deviation from the arrangement of the character rectangles, the error can be reduced by excluding these rectangles from the tilt detection processing target.

Further, the embodiment can be modified so that only the small circumscribing rectangle on the rightmost wing of the block among the circumscribing rectangles much smaller than the reference value is excluded from the processing target. That is, in FIG. 4, the circumscribing rectangle 404 is a circumscribing rectangle that is much smaller than the reference value, but is not a period, a comma, or the like, but a part of a character. Such a circumscribed rectangle appears in the characters "i", "j", etc. However, when the circumscribed rectangle 404 is excluded, the circumscribed rectangle 405 is located at the lower side, so an error occurs. Therefore, only the circumscribing rectangle with the smallest rightmost wing of a block having a high probability of a period or a comma is excluded from the processing target of inclination detection.

Further, depending on the font and the character size, there is a case where the heights of uppercase letters and lowercase letters do not differ as much as the threshold value. Even in such a case, the size shift appears as an error, so the circumscribed rectangle of the leftmost wing corresponding to the first character of the block is likely to be capitalized,
It is possible to further change the embodiment so that the conversion processing is always performed to the standard height.

[0019]

As described above, according to the present invention, the inclination of the image is obtained from the character string determination area.
The amount of processing is greatly reduced compared to the conventional method, and when the character area is extracted by the blurring processing, the inclination of the image is calculated with respect to the extracted area.
The inclination detection process and the region extraction process have good consistency. Also,
Even in English documents, the inclination of the image can be detected correctly.

[Brief description of drawings]

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

FIG. 2 shows a processing flowchart of the present invention.

FIG. 3 is a diagram showing a character string determination area whose horizontal length is three times or more the vertical length.

FIG. 4 is a diagram showing an example of an English character string in which the size and position of a circumscribing rectangle vary.

FIG. 5 is a diagram showing a height histogram of a circumscribed rectangle.

FIG. 6 is a diagram for converting the size of a circumscribed rectangle of an original image into a reference value.

[Explanation of symbols]

 Reference Signs List 101 image input unit 102 image reduction unit 103 block generation / classification unit 104 tilt detection unit 105 data storage unit 106 control unit

Claims (1)

Claim: What is claimed is: 1. An image tilt detection method for detecting a tilt of a character string region from a binarized document image, the step of reducing an input image, and a blurring process for the reduced image. A step of generating a block by subjecting the block to classification of a character string area from the generated block, and a step of detecting an image inclination from the position of a circumscribed rectangle of a predetermined area in the character string area. Image tilt detection method.