JP3712825B2 - Image processing method, apparatus, and recording medium - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to an image processing method, apparatus, and recording medium for accurately detecting the tilt of a document.
[0002]
[Prior art]
Conventionally, in character recognition processing and the like, a document image may be input with an inclination, but if the inclination is slight, recognition can be executed without correcting the inclination. However, when the inclination of the image is large, it is necessary to correct the inclination as preprocessing for recognition.
[0003]
As a method for detecting the inclination of the document image, the present applicant has previously proposed several methods. One of the methods is a method described in Japanese Patent Laid-Open No. 7-141465, and for the circumscribed rectangle of a character string, the inclination of the document image and the certainty of the inclination are obtained from the histogram of the inclination of the straight line connecting the neighboring rectangles. .
[0004]
Another method is an image inclination detection method and a table processing method described in Japanese Patent Application Laid-Open No. 7-105310. In this method, a circumscribed rectangle of a ruled line is obtained, and the inclination is determined from the arrangement of black runs in the rectangle. Calculate
[0005]
[Problems to be solved by the invention]
In the first method described above, the rectangle information of the character string is extracted based on the case where there is no inclination. Therefore, when the inclination of the inclination detection target document is large, it is difficult to appropriately extract the circumscribed rectangle of the character. there were. In this case, it can be determined from the extracted rectangular feature that the document is greatly inclined, but the reliability of the calculated inclination may be lowered. Similarly, the above-described second method does not necessarily have high reliability of the slope calculated when the slope is large.
[0006]
An object of the present invention is to extract high-accuracy by re-extracting character string and ruled line rectangle information on the basis of the already obtained rough inclination when it is determined that the inclination of the inclination detection target document is large. An object of the present invention is to provide an image processing method, apparatus, and recording medium capable of obtaining rectangular information and a highly reliable tilt detection value.
[0007]
[Means for Solving the Problems]
In order to achieve the above object, according to the first aspect of the present invention, a feature of a document component in a document image is extracted, a tilt of the document image is calculated based on the feature, and the calculated tilt is predetermined. When the calculated inclination is less than or equal to a predetermined threshold value, the calculated inclination is determined as the inclination of the document image, and the calculated inclination exceeds the predetermined threshold value The feature is extracted again from the document image based on the tilt, and the tilt of the document image is calculated based on the re-extracted feature.
[0008]
The invention according to claim 2 is characterized in that a circumscribed rectangle of a character string is used as a feature of the document component.
[0009]
According to the third aspect of the present invention, means for extracting the characteristics of the document constituent element in the document image, means for calculating the inclination of the document image based on the characteristics, and the calculated inclination exceeds a predetermined threshold value Means for determining whether or not the calculated inclination is equal to or less than a predetermined threshold , and means for determining the calculated inclination as an inclination of the document image; and when the calculated inclination exceeds a predetermined threshold , And means for re-extracting features from the document image based on the inclination, and means for calculating the inclination of the document image based on the re-extracted features.
[0010]
The invention described in claim 4 is characterized in that a circumscribed rectangle of a character string is used as a feature of the document component.
[0011]
The invention described in claim 5 is a computer-readable recording medium storing a program for realizing the image processing method described in claim 1 or 2 on a computer.
[0012]
DETAILED DESCRIPTION OF THE INVENTION
Hereinafter, an embodiment of the present invention will be described in detail with reference to the drawings.
FIG. 1 shows the configuration of an embodiment of the present invention. In the figure, 101 is an image input means, 102 is a document component position feature extraction means, 103 is an inclination determination means, 104 is a large inclination determination means, 105 is a data storage section, 106 is a control section for each process, and 107 is data. A transfer path and a control communication path.
[0013]
FIG. 2 is a process flowchart of the present invention. The processing operation of the present invention will be described below with reference to FIG.
First, a document image is obtained by the image input means 101 (step 201). The image input means may be a scanner, a fax machine, or a document image that has already been electronically filed.
[0014]
Next, the position features of the document components such as ruled lines and character strings in the document are extracted by the document component location feature extraction means 102 (step 202). FIG. 3 shows an example of extracting the position feature of the document component. In this example, circumscribed rectangle information of a character string is used as the position feature of the document component. First, the circumscribed rectangle of the character string is extracted. In FIG. 3, 301 is a character string in the document, and 302 is a circumscribed rectangle of the extracted character string. As a method for extracting the circumscribed rectangle, for example, a known technique (a method for extracting a connected component of black pixels and obtaining the circumscribed rectangle) described in the above-mentioned publication (Japanese Patent Laid-Open No. 7-141465) may be used.
[0015]
Next, the inclination determination unit 103 detects the inclination of the document image (step 203). As this inclination detection method, a known technique described in the above-mentioned publication (Japanese Patent Laid-Open No. 7-141465) may be used. That is, a circumscribed rectangle of a black pixel connected component is obtained, a rectangle corresponding to a character string is determined from the circumscribed rectangle, a plurality of reference points are set in each determined rectangle, and a straight line connecting between reference points of neighboring rectangles is set. The inclination of the image is detected based on the inclination histogram.
[0016]
Then, the large inclination determination unit 104 determines whether or not the detected inclination is large (step 204). Whether or not the inclination is large is determined in advance by determining a reference angle as a threshold value and comparing it with the threshold value. If the result of determination in step 204 is that the tilt is not large, the calculated angle is determined as the tilt of the document, and the process ends.
[0017]
On the other hand, when it is determined in step 204 that the inclination is large, the angle obtained by the method described in the above publication generally reflects the inclination of the document, but its reliability is not so high. . This is because the circumscribed rectangle is not sufficiently extracted as shown in FIG. That is, in the above-described method, the rectangle extraction is performed with the inclination 0 as a reference, and therefore the rectangle extraction is not necessarily accurately performed when the inclination is large.
[0018]
Therefore, in the present invention, the position characteristic of the document constituent element (such as a circumscribing rectangle of the character string) is obtained by the position characteristic extraction means 102 of the document constituent element by using the angle that largely reflects the inclination of the document obtained in step 203. Extract again (step 205). A circumscribed rectangle of the re-extracted character string is shown in FIG. In the figure, 401 is a character string in the document, and 402 indicates a circumscribed rectangle of the extracted character string. Based on the re-extracted feature, the inclination determination unit 103 recalculates the inclination (step 206).
[0019]
In the above-described embodiment, the circumscribed rectangle of the character string is used as the position feature of the document component, but the circumscribed rectangle of the ruled line can be used as the position feature. Similarly, when the circumscribed rectangle of the ruled line is used, when it is determined that the inclination is large, the circumscribed rectangle of the ruled line is re-extracted and the inclination is recalculated, thereby obtaining an accurate inclination.
[0020]
The present invention is not limited to the above-described embodiments, and can be realized by software. When the present invention is realized by software, as shown in FIG. 5, a general-purpose processing device including a CPU, a ROM, a RAM, a display device, a hard disk, a keyboard, a CD-ROM drive, a scanner, etc. is prepared. A program that realizes the image processing function of the present invention is recorded on a computer-readable recording medium such as a ROM. The document image input from the scanner is temporarily stored in a hard disk or the like. When the program is started, the temporarily stored document image data is read, image processing is executed, and the inclination detection result is passed to a character recognition processing program or the like.
[0021]
【The invention's effect】
As described above, according to the present invention, when it is determined that the inclination of the document image exceeds the predetermined threshold, the feature of the document component is extracted again, and the inclination is obtained based on the re-extracted feature. Therefore, it is possible to accurately detect the inclination of a document having a large inclination.
[Brief description of the drawings]
FIG. 1 shows a configuration of an embodiment of the present invention.
FIG. 2 shows a processing flowchart of the present invention.
FIG. 3 shows an example of extracting a position feature of a document component.
FIG. 4 shows an example in which the position features of document components are extracted again.
FIG. 5 shows a configuration example when the present invention is realized by software.
[Explanation of symbols]
DESCRIPTION OF SYMBOLS 101 Image input means 102 Position feature extraction means 103 of document component 103 Inclination determination means 104 Large inclination determination means 105 Data storage part 106 Control part 107 Data transfer path and control communication path

Claims (5)

The feature of the document component in the document image is extracted, the tilt of the document image is calculated based on the feature, it is determined whether the calculated tilt exceeds a predetermined threshold, and the calculated tilt When the calculated inclination is equal to or less than a predetermined threshold, the calculated inclination is determined as the inclination of the document image, and when the calculated inclination exceeds a predetermined threshold, the feature is extracted again from the document image based on the inclination, An image processing method characterized in that the inclination of the document image is calculated based on the re-extracted features.   The image processing method according to claim 1, wherein a circumscribed rectangle of a character string is used as a feature of the document component. Means for extracting a feature of a document component in a document image, means for calculating a tilt of the document image based on the feature, and means for determining whether the calculated tilt exceeds a predetermined threshold value A means for determining the calculated inclination as an inclination of a document image when the calculated inclination is equal to or less than a predetermined threshold; and the document based on the inclination when the calculated inclination exceeds a predetermined threshold. An image processing apparatus comprising: means for extracting a feature from an image again; and means for calculating an inclination of the document image based on the re-extracted feature. 4. The image processing apparatus according to claim 3, wherein a circumscribed rectangle of a character string is used as a feature of the document component. A computer-readable recording medium on which a program for realizing the image processing method according to claim 1 on a computer is recorded.

Images