JP2004145517A - Image correcting apparatus - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image correcting apparatus for accurately separating character rows, as well as simultaneously correcting inclination and meandering of the character rows. <P>SOLUTION: The apparatus comprises a storage means for storing image information; an expanded row generating means for generating an image including a plurality of expanded rows, by expanding to one direction the image stored in the storage means; a starting position detecting means for detecting the starting positions of the expanded rows by pixel strings in orthogonal directions perpendicular to the one direction; a row-separating means for separating to an orthogonal direction a range, in which a row to be corrected is structured; and a moving calculating means for moving the image stored in the storage means so as to horizontally align the starting positions of the expanded rows, based on the information on the starting positions of the expanded row within the range, in which the row to be corrected is structured. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image correction apparatus that corrects inclination and meandering of character lines and the like that occur in an image of a document such as a document photographed by a hand scanner or the like.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, various techniques have been proposed for capturing a document such as a document by a scanner or the like, and performing OCR (Optical Character Recognition) processing using the captured image to perform character recognition.
[0003]
Particularly, in an apparatus in which a user traces an image on a document using a relatively small scanner such as a hand scanner, the scanning direction at the time of photographing may be changed depending on the usage of the user. It is difficult to scan in a fixed direction with respect to the arrangement direction, and the resulting image may have a tilt or meandering compared to the original. As a result, when the degree of inclination or meandering is large, there is a problem that characters cannot be cut out accurately and the character recognition rate decreases.
[0004]
As a method of correcting the inclination of such an image, for example, a character image is binarized by binarizing the luminance value of each pixel as image data composed of a set of pixels arranged two-dimensionally. Scanning is performed by setting a large number of scanning lines parallel to each other for each pixel as binary image data, and data representing a character image of the binary image data is accumulated for each scanning line, and the accumulated value For example, a method has been proposed in which a projection profile is created by calculating a distribution in a direction orthogonal to a scanning line, and a rotation correction amount is calculated based on a variance value of the projection profile (for example, see Patent Document 1).
[0005]
[Patent Document 1]
Japanese Patent No. 3108979
[Problems to be solved by the invention]
However, in the above-described image inclination correction method, when creating a projection profile, the projection profiles for each character line overlap, and it is difficult to separate each character line. However, there is a problem that the operation may not be performed properly.
[0007]
Furthermore, since it is not possible to correct meandering for each character element constituting a character line, it is necessary to separately perform meandering correction on the character line.
[0008]
The present invention has been made in view of the above problems, and has as its object to provide an image correction apparatus capable of separating character lines with high accuracy and simultaneously correcting the inclination and meandering of the character lines.
[0009]
[Means for Solving the Problems]
An image correction apparatus according to the present invention includes a storage unit that stores image information, an expansion line generation unit that expands an image stored in the storage unit in one direction to generate an image including a plurality of expansion lines, and a plurality of expansion units. A row to be corrected is configured based on a start position detection unit that detects a start position of an expanded row, and a start position of the expanded row, for each pixel column in an image including rows in an orthogonal direction orthogonal to one direction. A row separating means for separating the range in the orthogonal direction and a storage means for storing the expanded rows in a horizontal manner based on the information on the start positions of the expanded rows in the range in which the rows to be corrected are formed. And a movement calculation unit that moves the selected image.
[0010]
With such a configuration, the range of the lower end position forming the character line is determined by detecting the start position of the inflated line and grouping the start position. Thus, there is little possibility that character lines overlap each other, and character lines can be separated with high accuracy.
[0011]
For this reason, even when a character line is photographed with a slight inclination, it is possible to simultaneously correct the inclination together with the meandering correction.
[0012]
Furthermore, since the image stored in the storage means is a binarized image whose luminance value is represented by a value of 0 or 1, the amount of memory used can be reduced and processing can be performed quickly. Since the load on the calculation unit can be reduced, it can be easily mounted on a portable information device or the like.
[0013]
Further, the detection of the start position of the expansion row is performed for each pixel column in the orthogonal direction by detecting the luminance value of the pixel of interest while moving the pixel of interest in the orthogonal direction. In the case where the pixels are consecutively arranged, the position where the pixel having the first brightness value of 0 is detected is set as the start position, so that it is possible to reduce the possibility that noise information due to dirt or the like is regarded as a character line. Appropriate image correction can be performed.
[0014]
Further, the detection of the start position of the dilation row is performed for a plurality of dilation rows, and based on the average value of the start position distributions of the plurality of dilation rows, for each pixel column in the orthogonal direction of the image stored in the storage means, , It is less likely to be affected by characters such as “j” and “p” that have jumped out below other characters, and more appropriate image correction can be performed.
[0015]
In addition, by configuring an information device equipped with the image correction device of the present invention, it is possible to capture a character or the like that has been subjected to image correction and is easy to perform character recognition and the like. In the information device, it is possible to increase the accuracy of the character reading.
[0016]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, embodiments of the present invention will be described with reference to the drawings.
[0017]
As an embodiment of the present invention, an image correction device of the present invention will be described.
[0018]
FIG. 1 is a flowchart illustrating a processing procedure of the image correction apparatus according to the embodiment of the present invention. FIG. 2 is a block diagram illustrating an example of a configuration of the image correction device according to the embodiment of the present invention.
[0019]
As shown in FIG. 2, an image correction device 30 according to an embodiment of the present invention is connected to an image input unit 1 that reads an original such as a character or a figure and takes an image of the original, and is connected to the image input unit 1. A CPU 2 which is an arithmetic unit for performing various processes described below based on an image photographed by the image input unit 1, and stores an image photographed by the image input unit 1 as image information of two-dimensionally arranged pixels. The first storage means 3 which further performs various arithmetic processing by the CPU 2 and the second storage means 4 which stores and stores an image obtained, and is further connected to the CPU 2 to obtain various kinds of calculated result information and the like. It comprises a display unit 5 for displaying necessary information to the user.
[0020]
In the present invention, the image information of a pixel refers to various types of information such as luminance information, color information, and density information of a pixel forming an image. In the present embodiment, luminance information of a pixel is used.
[0021]
As the image input unit 1, an arbitrary device selected from devices used for a known hand scanner or the like such as an optical device can be used.
[0022]
Each of the first storage unit 3 and the second storage unit 4 is a frame memory, and a known medium such as a flash memory can be used as a storage medium.
[0023]
The display unit 5 can be arbitrarily selected from a known display device, for example, a known device such as an LCD (Liquid Crystal Display), an EL (Electro-Luminescent), and a CRT (Cathode Ray Tube).
[0024]
Next, processing steps when the image correction apparatus according to the embodiment of the present invention performs image correction will be described with reference to FIG.
[0025]
First, an image (hereinafter, referred to as an original image) 10 captured by the image input unit 1 is developed as luminance information which is image information of pixels arranged two-dimensionally in the first storage unit 3 via the CPU 2. (Step S10).
[0026]
An example of the original image 10 is shown in FIG. The original image 10 shown in FIG. 3 is an image obtained by synthesizing the partial images photographed by the hand scanner, and the trajectory when tracing over the original using the hand scanner is shifted with respect to the direction of the character line. Therefore, the original image 10 shown in FIG. 3 has meandered.
[0027]
In the present embodiment, as described above, the original image 10 is a set of pixels arranged two-dimensionally, and is a monochrome image in which the luminance value of each pixel is a multi-level (for example, 256 gradations). is there.
[0028]
Further, in the present embodiment, a 256 × 16 pixel CCD is used as a photographing unit of the image input unit 1, and a first storage unit 3 for developing an image is horizontal with respect to the plane of FIG. A frame memory of (direction (horizontal) × vertical direction (vertical) with respect to the paper surface) = 1000 × 400 pixels was used.
[0029]
Next, the CPU 2 uses a known method to set either 0 (black) or 1 (white) for each pixel to the original image 10 stored in the first storage unit 3 as luminance information. The stored binarization processing is performed (step S11). By performing this binarization processing, it is possible to reduce the amount of memory used and to speed up the processing, and to reduce the load on the CPU 2.
[0030]
FIG. 4 shows an example of the binarized image 11 that has been subjected to such binarization processing. Pixels that appear black are pixels in which a luminance value “0” is stored as luminance information, and pixels that appear white are pixels in which a luminance value “1” is stored as luminance information. In the present invention, any known method can be used for the method of binarizing the image, and there is no limitation.
[0031]
Returning to FIG. 1, next, the CPU 2 performs horizontal expansion processing on the binary image 11 stored in the first storage means 3 to expand the black pixel portion of the image in the horizontal direction ( Step S12).
[0032]
This horizontal expansion processing will be further described. FIG. 5 is a flowchart showing the steps of the horizontal dilation process (dilation row generation means).
[0033]
In FIG. 5, first, the CPU 2 sets an arbitrary pixel in the first storage unit 3 as a target pixel. In practice, the pixel located at the end of the binarized image 11, that is, the pixel located at the origin O at the upper left on the paper surface of FIG. Then, the luminance value of the target pixel is detected (step S21).
[0034]
Subsequently, the CPU 2 determines whether or not the luminance value of the pixel of interest is 0 (black) (step S22). If the luminance value is 0 (black), the corresponding value in the second storage unit 4 is determined. The luminance value of a predetermined range of pixels before and after the target pixel in the processing direction is set to 0 (black) (step S23). On the other hand, when the luminance value of the target pixel is not 0 (is 1), this processing is not performed.
[0035]
Step S23 will be described with reference to FIG. In FIG. 6, when the luminance value of a certain pixel of interest A in the first storage means 3 is 0 (black), a pixel within a predetermined range before and after in the processing direction (in this embodiment, 20 pixels before and after). Regarding each pixel (a total of 41 pixels including the pixel of interest), the luminance value of the corresponding pixel in the second storage means 4 is uniformly set to 0 (black) regardless of the original luminance value. The predetermined range is preferably about 20 pixels before and after practically.
[0036]
Returning to FIG. 5, subsequently, the CPU 2 determines whether or not processing has been completed for all pixels constituting the first storage means 3 (step S24). Ends the processing. On the other hand, if the processing has not been completed for all the pixels, the target pixel is moved in the processing direction in FIG. 4 (step S25), and the process returns to the step of detecting the luminance value of the target pixel (step S21).
[0037]
By performing such a process, the horizontally expanded image 13 as shown in FIG. 7 can be obtained on the second storage unit 4. FIG. 7 is a diagram illustrating an example of the horizontal dilated image 13 in the present embodiment.
[0038]
As shown in FIG. 7, the horizontal dilated image 13 is composed of a plurality of dilated character lines. Here, the expanded character line (expanded line) means a pixel having a luminance value of 0, that is, a continuous block of black pixels. In the present embodiment, the four expanded character lines LA to LD of the horizontal expanded image 13 correspond to the four character lines A to D in the binarized image 11, respectively.
[0039]
Returning to FIG. 1, after the horizontal dilation process (step S12), the CPU 2 detects the start position of the black pixel in the vertical direction for each dilated character line (step S13). In other words, the CPU 2 starts from the leftmost vertical pixel column (referred to as a series of pixels in the vertical direction (column direction) in the image) of the horizontal expanded image 13 in FIG. When scanning is performed in the Y-axis direction from the bottom side and the continuation of black pixels (pixels with a luminance value of 0) continues more than a predetermined number, the position of the pixel where the black pixel is detected for the first time (hereinafter, referred to as (Referred to as a start position) is detected for each expanded character line. That is, in the horizontally expanded image 13 of the present embodiment, the starting positions of the four expanded character lines LA to LD are detected for each vertical pixel column.
[0040]
A method for detecting the expanded character line will be described with reference to FIG. FIG. 8 is a flowchart showing the steps of the expanded character line start position detection processing (start position detection means) according to the embodiment of the present invention.
[0041]
In FIG. 8, first, the CPU 2 sets a pixel of interest for an arbitrary vertical pixel column of the horizontal expanded image 13 stored in the second storage unit 4, and moves the pixel of interest to obtain the luminance of each pixel of interest. The value is detected (step S31). Note that the setting of the target pixel actually starts from the lowermost pixel in FIG.
[0042]
Next, the CPU 2 determines whether or not the black pixels are continuously detected for a predetermined number or more (step S32). If the black pixels are continuously detected for the predetermined number or more, the black pixels are started first. The position of the pixel is stored as the start position of the expanded character line (step S34). On the other hand, if the continuation of black pixels is less than the predetermined number, the continuation is regarded as noise information, not an expanded character line, and the pixel of interest is moved (step S36) to start the next expanded character line. Proceed to position detection processing. Note that, in the present embodiment, it is desirable that this predetermined number is set to around 20 for practical use.
[0043]
Then, it is determined whether or not the target pixel has reached the upper end of the vertical pixel column (step S35), and if it has reached, the process is terminated.
[0044]
The above-described processing is performed on all the vertical pixel rows (the entire screen) constituting the horizontal dilated image 13. By performing such processing, a series of short black pixels is regarded as noise information. Therefore, it is possible to extract and process only information of an expanded character line composed of character lines, and to perform processing by the influence of noise information. A configuration that is less susceptible to noise can be realized.
[0045]
As described above, as a result of performing the expanded character line start position detection processing on all the vertical pixel rows, the obtained expanded position of the expanded character line is obtained with respect to the vertical position and the start position at each vertical position. FIG. 9 shows a result (histogram) obtained by plotting the accumulated value obtained by accumulating the numbers of. FIG. 9 shows the distribution of expanded character line start points in the embodiment of the present invention, and shows the range of the lower end position of the character line.
[0046]
Returning to FIG. 1, the CPU 2 determines the range of the lower end position of each character line (the range W in FIG. 9 as an example) using the relationship shown in FIG. This processing (row separation means) is obtained by detecting, from the histogram shown in FIG. 9, the width of a mountain having an area equal to or greater than a predetermined value (hereinafter, this processing is referred to as grouping) (step S14). In the present embodiment, as shown in FIG. 9, there are four peaks, which respectively correspond to the range of the start positions of the expanded character lines LA to LD.
[0047]
In this step S14, if the area of the peak of the histogram is smaller than a predetermined value, the information is regarded as noise information and the information of the integrated value is ignored. With such a configuration, it is possible to detect an appropriate expanded character line start position with less influence of noise information.
[0048]
FIG. 10 is a diagram showing the distribution of the start positions of each expanded character line by grouping the start positions of the expanded character lines in the embodiment of the present invention. The horizontal axis indicates the horizontal position of the dilated image in FIG. 7, and the vertical axis indicates the vertical position. As described above, the distribution of the start positions, that is, the lower end positions of the four expanded character lines LD, LC, LB, and LA is shown in order from the top.
[0049]
As described above, according to the configuration of the present invention, by performing the grouping of the starting position of the expanded character line, the range of the lower end position of the character does not overlap and is easily separated, so that the character line can be separated with high accuracy. .
[0050]
Returning to FIG. 1, next, the CPU 2 calculates a shift amount to be corrected in the vertical direction for each of the four expanded character lines LA to LD in FIG. 10 for all the vertical pixel rows (step S15). Then, the average of the shift amounts of the four expanded character lines LA to LD is calculated (step S16).
[0051]
FIG. 11 is a diagram showing a relative shift amount distribution of each vertical pixel column in the embodiment of the present invention. The horizontal axis shows the horizontal position, and the vertical axis shows the relative deviation calculated by averaging all the character lines. From the relationship shown in FIG. 11, the shift amount to be corrected in the vertical direction is determined for each vertical pixel column.
[0052]
In FIG. 1, the CPU 2 (movement calculation unit) moves each vertical pixel column of the binarized image 11 stored in the first storage means 3 in the vertical direction by the displacement amount based on the average value of the displacement amount. To end the processing (step S17).
[0053]
FIG. 12 shows an example of the corrected image 15 as a result of performing image correction on the original image 10 by the image correction device 30 according to the embodiment of the present invention.
[0054]
As can be seen from FIG. 12, compared to the original image 10, both the inclination of the character line and the degree of meandering are corrected. As described above, since the inclination and meandering are corrected by using the image correction device of the present invention, it is possible to suppress a reduction in the recognition rate in the subsequent processing such as OCR.
[0055]
According to the image correction apparatus of the present invention, the shift amount to be corrected uses the average value of the shift amounts calculated for each character line. Therefore, for example, in a certain character line, for example, “j”, “p”, etc. Even if there is a character element that protrudes downward, it is possible to realize a configuration in which a line is less affected by the character element.
[0056]
Note that the image correction device of the present invention is not limited to the configuration shown in the present embodiment. For example, based on the histogram of the start position of the expanded character line shown in FIG. 9, all vertical pixels forming the image are referred to by referring only to the shift amount to be corrected calculated for the character line including the largest integrated value. A configuration for performing correction on a column may be used. This means that the character line containing the largest integrated value usually means the longest character line, so that by detecting the amount of deviation to be corrected for that character line and performing correction, most of the entire image is corrected. This is because it can be corrected. Practically, even with such a configuration, it is possible to satisfactorily correct the tilt and meander of the image. In addition, the processing can be performed more quickly than in the case where the start positions are calculated for all the character lines constituting the image and the deviation is corrected based on the average value.
[0057]
Using the image correction device 30 according to the embodiment of the present invention, image correction was performed using 20 random business cards as samples. The number of characters provided for OCR is 390 for telephone number, and 1026 for mail address and URL. After the image correction, the CPU 2 performs an OCR process based on the image stored in the first storage unit 3 and calculates the proportion of characters correctly recognized as a result. As a result, the telephone number in the business card, the mail address, In the entire URL, the normal reading rate can be improved by about 25% by the effect of the meandering correction and the inclination correction when the image correction of the present invention is performed, compared to the normal reading rate without correction. .
[0058]
In the embodiment of the present invention, although the function of the display unit 5 is not particularly described, an acquired original image, a binarized image, or the like is displayed to the user, an error message or the like is displayed. If a message or the like of necessary input contents is displayed, a device structure that is more user-friendly can be realized.
[0059]
Further, by using the image correction device of the present invention, the amount of memory used can be reduced, processing can be performed quickly, and the load on the arithmetic unit can be reduced, so that it can be easily mounted on portable information devices and the like. . FIG. 13 shows an example in which such an image correction device 30 is mounted on an information device such as a mobile phone device.
[0060]
FIGS. 13A and 13B are external views of a mobile phone device 50 equipped with the image correction device 30, wherein FIG. 13A is a front view thereof, and FIG. 13B is a perspective view of a main part showing an end portion on which the image correction device 30 is mounted. is there.
[0061]
The mobile phone device 50 has a configuration in which the image correction device 30 is built in a known mobile phone device such as an antenna unit 51, a speaker unit 52, a display unit 5 such as an LCD, a key unit 54, and a microphone unit 55.
[0062]
By mounting the image correction device of the present invention, the surface of the image input unit 1 of the image correction device 30 for reading information represented by light and shade such as characters and figures is provided on the lower surface of the mobile phone device 50. Thus, it is possible to provide the mobile phone device 50 with the built-in image correction device 30 which is very easy for the user to use.
[0063]
By using a mobile phone equipped with such an image correction device, document information such as a URL and a two-dimensional barcode can be read and used for processing such as OCR. An apparatus can be provided.
[0064]
Here, the information device is not limited to the above-mentioned portable telephone device, but includes various known information devices such as a small personal computer such as a PDA and a notebook, and a PDA (personal digital assistant). Nor.
[0065]
In the present embodiment, an example has been shown in which the inclination and meandering of each character element (such as characters and numbers) on an image read by the image correction device are corrected, but the image correction device of the present invention can correct the inclination and meandering. The image is not limited to this. For example, the image correction apparatus of the present invention corrects the inclination and correction of a read image of a document in which information such as barcodes and graphics are arranged in one direction on a screen instead of or in addition to character elements. It goes without saying that you can do it.
[0066]
【The invention's effect】
As described above, by using the image correction device of the present invention, it is possible to realize an image correction device that can simultaneously correct the inclination and meandering of a character line and can separate a character line with high accuracy.
[Brief description of the drawings]
FIG. 1 is a flowchart illustrating a processing procedure of an image correction device according to an embodiment of the present invention; FIG. 2 is a block diagram illustrating an example of a configuration of the image correction device according to an embodiment of the present invention; FIG. 4 is a diagram showing an example of an original image in the embodiment of the present invention. FIG. 4 is a diagram showing an example of a binarized image in the embodiment of the present invention. FIG. 5 is a flowchart showing steps of a horizontal dilation process in the embodiment of the present invention. FIG. 6 is a schematic diagram illustrating the contents of horizontal expansion processing according to the embodiment of the present invention. FIG. 7 is a diagram illustrating an example of a horizontal expansion image according to the embodiment of the present invention. FIG. 9 is a flowchart showing the steps of a process of detecting the start position of an expanded character line. FIG. 9 is a diagram showing a histogram of the start position of an expanded character line according to the embodiment of the present invention. FIG. 11 is a diagram showing a distribution of start positions of expanded character lines. FIG. 11 is a diagram showing a distribution of relative shift amounts of start positions of expanded character lines in the embodiment of the present invention. FIG. 13 (a) is a front view of a mobile phone device equipped with an image correction device according to the embodiment of the present invention, and FIG. 13 (b) is a front view of the mobile phone device equipped with the image correction device according to the embodiment of the present invention. Perspective view of the main part showing the cut end
1 image input unit 2 CPU (arithmetic unit)
3 First storage unit 4 Second storage unit 5 Display unit 10 Original image 11 Binary image 13 Horizontal expanded image 15 Corrected image 30 Image correction device 50 Mobile phone device 51 Antenna unit 52 Speaker unit 54 Key unit 55 Microphone section

Claims (5)

Storage means for storing image information;
An expansion line generation unit that expands the image stored in the storage unit in one direction to generate an image including a plurality of expansion lines,
A start position detection unit that detects a start position of the expansion row, for each pixel column in an image including the plurality of expansion rows in a direction orthogonal to the one direction,
Based on the start position of the expansion row, a row separation unit that separates the range in which the row to be corrected is configured in the orthogonal direction,
Movement for moving the image stored in the storage means such that the start positions of the expansion rows are horizontally aligned based on information on the start positions of the expansion rows within the range in which the rows to be corrected are configured. An image correction device, comprising: a calculation unit. The image correction apparatus according to claim 1, wherein the image stored in the storage unit is a binarized image whose luminance value is represented by a value of 0 or 1. The detection of the start position of the expansion row is performed for each pixel column in the orthogonal direction by detecting the luminance value of the pixel of interest while moving the pixel of interest in the orthogonal direction. 3. The image correction apparatus according to claim 2, wherein, when the number of consecutive pixels exceeds a predetermined value, a position where a pixel having a luminance value of 0 is detected first is set as the start position. Perform the detection of the start position of the expansion line for a plurality of expansion lines,
The method according to claim 1, wherein the image is moved in the orthogonal direction for each pixel column in the orthogonal direction of the image stored in the storage unit, based on an average value of a start position distribution of the plurality of expansion rows. The image correction device according to claim 3. An information device, comprising the image correction device according to any one of claims 1 to 4.

Images