KR101725501B1 - Method and apparatus for recognizing character - Google Patents
Method and apparatus for recognizing character Download PDFInfo
- Publication number
- KR101725501B1 KR101725501B1 KR1020160088383A KR20160088383A KR101725501B1 KR 101725501 B1 KR101725501 B1 KR 101725501B1 KR 1020160088383 A KR1020160088383 A KR 1020160088383A KR 20160088383 A KR20160088383 A KR 20160088383A KR 101725501 B1 KR101725501 B1 KR 101725501B1
- Authority
- KR
- South Korea
- Prior art keywords
- character
- center
- gravity
- straight line
- radial straight
- Prior art date
Links
Images
Classifications
-
- G06K9/3233—
-
- G06K9/3258—
-
- G06K9/3283—
-
- G06K9/4604—
-
- G06K2209/01—
Abstract
The present invention relates to a character recognition method and apparatus for extracting a feature point using a radial straight line inspection algorithm or a polygonal check algorithm in order to quickly recognize a character included in an image.
And extracting feature points according to one of a radial straight line inspection algorithm or a polygon check algorithm based on a reference point inside the ROI, wherein the radial straight line inspection algorithm includes a step of extracting a plurality of reference points from a reference point Wherein the polygon inspection algorithm includes a plurality of closed polygons whose diameters are constantly increased around the reference point, and wherein the closed polygons have a diameter gradually increasing from the center of the closed polygons, And extracts feature points based on the relationship of the characters.
Description
The present invention relates to a character recognition method and apparatus, and more particularly, to a character recognition method and apparatus for extracting characteristic points using a radial straight line inspection algorithm or a polygon inspection algorithm to quickly recognize characters included in an image.
Conventionally, character recognition is generally performed by converting a text file or an electronic document using a scanner. Recently, however, a method of recognizing a character displayed on an image photographed from a camera has been proposed.
First, Patent Document 1 (Japanese Patent Application Laid-Open No. 1998-065813) discloses a technology for dividing a binary input character into small regions with a real length and recognizing the characters in a small region in a statistical manner.
In addition, Japanese Patent Application Laid-Open No. 10-0667156 (hereinafter referred to as "Patent Document 2") discloses a technique of selecting a character region in a character image acquired using a portable camera to find a virtual character size, And recognizes the character after locally binarizing it.
However, in the case of Patent Document 1, the binarized input character is merely a character segmentation by dividing the input character into a real number in length, and in the case of Patent Document 2, the character extraction is performed by performing localization within the extracted virtual character region And Patent Documents 1 and 2 do not restrict input characters, so that it takes a long processing time to recognize a character of an image and a problem of occupying a large amount of memory.
The inventor of the present invention has made efforts to research for a long time in accordance with the user demand as described above, and thus the present invention has been completed.
The object of the present invention is to extract feature points according to one of a radial straight line inspection algorithm or a polygon check algorithm in order to quickly recognize a character in an image, simplify a character group based on the rotation rate, It is possible to provide a character recognition method and apparatus with improved processing speed by applying a recognition algorithm.
The character recognition method of the present invention includes a step of setting a region of interest of an image and a step of extracting feature points according to either a radial straight line inspection algorithm or a polygonal check algorithm based on a reference point inside the ROI, Wherein the straight line inspection algorithm extracts feature points based on a relationship between the radial straight line and a character, the polygon inspection algorithm comprising a plurality of closed polygons with a constant diameter increasing around the reference point, And extracts feature points based on the relationship between the closed polygon and the character.
A character recognition method in which a group-by-group character recognition algorithm is sequentially applied based on a turnover rate comprises: setting a region of interest of the image; extracting feature points according to a center-of-gravity standard deviation checking algorithm based on a center- And outputting the character if the character is the first group and extracting the feature point according to one of the radial straight line inspection algorithm and the polygon check algorithm if the character is the second group .
The character recognition apparatus detects edges of characters in an image to extract edges and binds the pixels determined as the boundaries to form a point of interest located at the upper left corner and a point located at the lower right corner of the rectangle as vertexes of the rectangle A plurality of radial straight lines are arranged on the basis of the reference points of the ROIs and are calculated on the basis of the relationship between the radial straight lines and characters or the diameter is constantly increased around the reference point of the ROI And a feature point extracting unit that extracts feature points based on the relationship between the closed polygon and the character.
According to the present invention, the present invention provides a method for recognizing characters from a received image quickly. In particular, a character group based on the rotation rate can be set, and a character group having no influence on the rotation rate and a character group having an influence on the rotation rate can be distinguished and can be quickly recognized.
According to the present invention, there is an effect that the accuracy of character recognition is improved by being less influenced by the rotation of the character, the size change, and the viewpoint change in the image.
1 is a flowchart for explaining a character recognition method according to a first embodiment of the present invention.
2 is a flowchart illustrating a character recognition method according to a second embodiment of the present invention.
3 is a diagram for explaining a character recognition method using a center-of-gravity standard deviation inspection algorithm according to the present invention.
4 is a diagram for explaining a character recognition method using the radial straight line inspection algorithm of the present invention.
5 is a diagram for explaining a character recognition method using the polygon inspection algorithm of the present invention.
6 is a block diagram for explaining the character recognition apparatus of the present invention.
It is to be understood that the specific structural or functional description of embodiments of the present invention disclosed herein is for illustrative purposes only and is not intended to limit the scope of the inventive concept But may be embodied in many different forms and is not limited to the embodiments set forth herein.
The embodiments according to the concept of the present invention can make various changes and can take various forms, so that the embodiments are illustrated in the drawings and described in detail herein. It is not intended to be exhaustive or to limit the invention to the particular forms disclosed, but on the contrary, is intended to cover all modifications, equivalents, or alternatives falling within the spirit and scope of the invention.
The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention. The singular expressions include plural expressions unless the context clearly dictates otherwise. In this specification, the terms "comprises" or "having" and the like are used to specify that there are features, numbers, steps, operations, elements, parts or combinations thereof described herein, But do not preclude the presence or addition of one or more other features, integers, steps, operations, components, parts, or combinations thereof.
Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings attached hereto.
1 is a flowchart for explaining a character recognition method according to a first embodiment of the present invention.
Referring to FIG. 1, the character recognition method of the present invention can pre-process a received image (S110). The image preprocessing step is to reduce the degree of shaking when the area of interest is visualized. The image preprocessing step includes a warping step of changing the perspective of the image, a blurring step of removing noise of the image, Accumulating step, binarizing the image, and morphological computing step.
Thereafter, a region of interest among the preprocessed images can be set (S120). In the interest region setting step, first, a boundary detection is performed to extract a corner corresponding to a boundary between a white portion and a black portion. The pixels determined as the boundary are grouped together, and a point located at the upper left corner and a portion located at the lower right corner You can set the region of interest by making the point a rectangle's vertex. Subsequently, the model enclosed by the rectangle is moved to the square space, and the data of the square shape is substituted without changing the angle of the character.
Subsequently, feature points may be extracted according to either the radiographic straight line inspection algorithm or the polygon check algorithm based on the reference points inside the ROI (S130). The reference point is not limited to a center point at any point in the ROI.
The radial straight line inspection algorithm may extract a plurality of radial straight lines from the reference point and extract feature points based on the relationship between the radial straight line and the character. The polygon inspection algorithm can arrange a plurality of closed polygons whose diameters are constantly increased around the reference point and extract feature points based on the relationship between the closed polygons and the characters.
The relationship between the radial straight line and the character may be at least one of a number of intersections of the radial straight line and the character, a distance between the radial straight line and the character, and a statistical distribution of the radial straight line and the character, The relationship of the characters may be at least one of the number of intersections of the closed polygon and the character, the distance of the closed polygon and the character, and the statistical distribution of the closed polygon and the character.
The radial straight line inspection algorithm and the polygon inspection algorithm may be selectively performed, and may be sequentially performed according to the embodiment.
The radial straight line inspection algorithm implements a matrix containing 2 n radiation lines and performs an AND operation with a square ROI. When 2n radial straight lines are used, the number of points obtained by the AND operation is returned as a data array capable of storing n values. The values of each of the data arrays are used as feature values to distinguish the characters. Such a method may be advantageous to determine the turnover rate of a group of characters whose result may vary depending on the turnover rate.
The polygon inspection algorithm can be calculated by arranging a closed polygon whose diameter increases constantly around the reference point. The feature points can be extracted by the number of intersections of the closed polygon and the character, the distance between the closed polygon and the character, and the statistical distribution of the closed polygon and the character. It is not necessary to calculate the area, and the number of pixels is counted so that the amount of computation does not increase greatly, and the extracted data is returned as an array, so that it can be used as the feature value. That is, the character can be recognized based on the extracted feature points through the radial straight line inspection algorithm and the polygon check algorithm.
2 is a flowchart illustrating a character recognition method according to a second embodiment of the present invention.
Referring to FIG. 2, the character recognition method of the present invention receives an image from a camera and preprocesses the image (S210). The image preprocessing step is to reduce the degree of shaking when the area of interest is visualized. The image preprocessing step includes a warping step of changing the perspective of the image, a blurring step of removing noise of the image, Accumulating step, binarizing the image, and morphological computing step.
The warping step can change the oblique image as seen from the front. If a value is specified assuming that the obliquely viewed image is seen through this step, the effect of resizing the image can be shown. Therefore, it is possible to reduce the amount of data and to show the effect of viewing the image from the front by changing the input image as seen from the far front.
The blurring step is a step of removing the noise of the image and adjusting the extraordinary pixel to be similar to the neighboring pixel. This step removes noise from the image. In the image accumulating step, the values of two or more images are multiplied by weights, and the noise can be maximized by accumulating the images. When the screen is stopped, the noise on the stopped screen can be canceled.
In the binarizing step, the values of the pixels exceeding a specific value are classified into a maximum value, and when the value is less than a specific value, a minimum value may be classified. The floor where the character is located is assumed to be white, and it can be filtered by three values of HSV to distinguish characters and bases, and characters can be displayed in black on a white background. The morphological operation step may perform an erosion operation that shapes the shape and an expansion operation that inflates the shape to fill gaps in the characters or eliminate unnecessary protrusions.
After the preprocessing step, a region of interest (ROI) may be set in the preprocessed image (S220). In the interest region setting step, edge detection is performed first, and edges corresponding to a boundary between a white portion and a black portion are extracted. It is possible to set the region of interest by connecting the pixels determined as the boundaries and setting the vertex located at the upper left corner and the vertex located at the lower right corner as rectangle vertices.
Subsequently, the model enclosed by the rectangle is moved to the square space, and the data of the square shape is substituted without changing the angle of the character. The center of gravity can be matched to the center of the square.
After setting the ROI, the feature point may be extracted according to the center-of-gravity standard deviation checking algorithm based on the center-of-gravity point within the ROI (S230).
The center-of-gravity standard deviation checking algorithm can extract the minutiae by calculating the standard deviation between the center-of-gravity point and the edge points of the character. And calculates a distance D i with respect to the center of gravity M (x, y) of the region of interest and all the border points (a i , b i ) in the region of interest (ROI). Then, the distance is calculated for all the edges, the distance is averaged, and the standard deviation between all the D i is obtained. The standard deviation values are used as feature values to distinguish the numbers. That is, the standard deviation can be obtained by the following equations (1) to (3).
[Formula 1]
[Formula 2]
[Formula 3]
After recognizing the shape of the character, it may be determined whether rotation rate calculation is necessary (S240). At this time, it is determined which one of the first group and the second group is set in advance.
For example, the first group can be divided into characters having no influence on the rotation rate such as' 2 ',' 3 ',' 4 ',' 5 ',' 7 ',' 8 ',' 0 ',' The second group can be divided into characters that affect turnover such as '6', '9', 'x', '+'. The letters of the first group and the letters of the second group represent numbers and symbols, but the present invention is not limited thereto.
If the type of the character corresponds to the first group, the character can be directly outputted (S260). If the character type corresponds to the second group, the feature point is extracted according to one of the radial straight line inspection algorithm and the polygon check algorithm (S250).
The radial straight line inspection algorithm implements a matrix containing 2 n radiation lines and performs an AND operation with a square ROI. When 2n radial straight lines are used, the number of points obtained by the AND operation is returned as a data array capable of storing n values. The values of each of the data arrays are used as feature values to distinguish the characters. Such a method may be advantageous to determine the turnover rate of a group of characters whose result may vary depending on the turnover rate.
The polygon inspection algorithm can be calculated by arranging a closed polygon whose diameter increases constantly around the reference point. The feature points can be extracted by the number of intersections of the closed polygon and the character, the distance between the closed polygon and the character, and the statistical distribution of the closed polygon and the character. It is not necessary to calculate the area, and the number of pixels is counted so that the amount of computation does not increase greatly, and the extracted data is returned as an array, so that it can be used as the feature value. Characters can be recognized based on the extracted feature points through the radial straight line inspection algorithm and the polygon check algorithm.
That is, a group based on the rotation rate can be set, and a group having no influence on the rotation rate and a group having an influence on the rotation rate can be distinguished and can be quickly recognized.
3 is a diagram for explaining a character recognition method using the standard deviation of the present invention.
Referring to FIG. 3, a method of extracting feature points and recognizing characters by calculating the standard deviation between the center of
A frame is searched for within the ROI and a distance D i is calculated with respect to all center points a i , b i and a center of gravity point M (x, y) in the ROI. Then, the distance is calculated for all the edges, the distance is averaged, and the standard deviation between all the D i is obtained. The standard deviation values are used as feature values to distinguish the numbers. That is, the standard deviation can be obtained by the following equation (3).
[Formula 1]
[Formula 2]
[Formula 3]
4 is a diagram for explaining a character recognition method using the radial straight line inspection algorithm of the present invention.
4, the radial
The radial straight line inspection algorithm may extract feature points based on a
5 is a diagram for explaining a character recognition method using the polygon inspection algorithm of the present invention. Referring to FIG. 5, the
6 is a block diagram for explaining the character recognition apparatus of the present invention.
6, the
The
The
The region of interest (ROI)
The feature
The
While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is evident that many alternatives, modifications and variations will be apparent to those skilled in the art. Accordingly, the true scope of the present invention should be determined by the technical idea of the appended claims.
600; Character recognition device
610; Image receiving unit
620; The image pre-
630; The region-
640; The feature point extracting unit
650; Recognition unit
660; Communication section
Claims (6)
(b) extracting feature points according to one of a radial straight line inspection algorithm or a polygon check algorithm based on a reference point inside the ROI,
Wherein the radial straight line inspection algorithm includes a plurality of radial straight lines arranged from a reference point, extracting feature points based on the relationship between the radial straight line and characters,
Wherein the polygon inspection algorithm includes arranging a plurality of closed polygons whose diameters are constantly increased around the reference point and extracting feature points based on the relationship between the closed polygons and the characters.
Wherein the relationship between the radial straight line and the character is at least one of a number of intersections of the radial straight line and a character, a distance between the radial straight line and a character, and a statistical distribution of the radial straight line and a character,
Wherein the relationship between the closed polygon and the character is at least one of a number of intersections of the closed polygon and the character, a distance between the closed polygon and the character, and a statistical distribution of the closed polygon and the character.
(a) setting a region of interest of an image;
(b) extracting feature points according to a center-of-gravity standard deviation inspection algorithm based on a center-of-gravity point within the ROI;
(c) determining whether the character turnover rate is calculated; And
(d) if the character is a first group, outputting the character, and if the character is a second group, extracting a feature point according to one of a radial straight line inspection algorithm and a polygon check algorithm.
Wherein the center-of-gravity standard deviation checking algorithm calculates the standard deviation between the center-of-gravity point and the boundary points of the character by substituting into the following equations (1), (2), and (3).
[Formula 1]
[Formula 2]
[Formula 3]
A plurality of radial straight lines are arranged based on a reference point of the region of interest and calculated based on a relationship between the radial straight line and a character,
And a feature point extracting unit that arranges a plurality of closed polygons whose diameters are constantly increased around the reference point of the ROI and extracts the feature points based on the relationship between the closed polygons and the characters.
Wherein the feature point extracting unit extracts feature points according to a center-of-gravity standard deviation checking algorithm based on a center-of-gravity center point within the ROI, and the center-of-gravity standard deviation checking algorithm calculates a standard deviation between the center- (1), (2), and (3).
[Formula 1]
[Formula 2]
[Formula 3]
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020160088383A KR101725501B1 (en) | 2016-07-13 | 2016-07-13 | Method and apparatus for recognizing character |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020160088383A KR101725501B1 (en) | 2016-07-13 | 2016-07-13 | Method and apparatus for recognizing character |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| KR101725501B1 true KR101725501B1 (en) | 2017-04-13 |
Family
ID=58579911
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| KR1020160088383A KR101725501B1 (en) | 2016-07-13 | 2016-07-13 | Method and apparatus for recognizing character |
Country Status (1)
| Country | Link |
|---|---|
| KR (1) | KR101725501B1 (en) |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101900903B1 (en) | 2017-09-19 | 2018-11-02 | 충북대학교 산학협력단 | Image orientation estimation method based on center of mass of partitioned images, character recognition apparatus and method using thereof |
| CN109376731A (en) * | 2018-08-24 | 2019-02-22 | 北京三快在线科技有限公司 | A kind of character recognition method and device |
| KR102152260B1 (en) | 2020-03-04 | 2020-09-04 | 주식회사 로민 | Apparatus and method for recognizing key-value relationship |
| US11678012B2 (en) | 2017-11-10 | 2023-06-13 | Samsung Electronics Co., Ltd. | Apparatus and method for user interest information generation |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0668304A (en) * | 1992-08-17 | 1994-03-11 | Fujitsu Ltd | Handwritten character normalization system |
| JPH0844824A (en) * | 1994-08-03 | 1996-02-16 | Mitsubishi Heavy Ind Ltd | Normalizing device |
| JP2001307022A (en) * | 2000-04-21 | 2001-11-02 | Oki Electric Ind Co Ltd | Character recognizing device and character recognizing method |
-
2016
- 2016-07-13 KR KR1020160088383A patent/KR101725501B1/en active IP Right Grant
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JPH0668304A (en) * | 1992-08-17 | 1994-03-11 | Fujitsu Ltd | Handwritten character normalization system |
| JPH0844824A (en) * | 1994-08-03 | 1996-02-16 | Mitsubishi Heavy Ind Ltd | Normalizing device |
| JP2001307022A (en) * | 2000-04-21 | 2001-11-02 | Oki Electric Ind Co Ltd | Character recognizing device and character recognizing method |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR101900903B1 (en) | 2017-09-19 | 2018-11-02 | 충북대학교 산학협력단 | Image orientation estimation method based on center of mass of partitioned images, character recognition apparatus and method using thereof |
| US11678012B2 (en) | 2017-11-10 | 2023-06-13 | Samsung Electronics Co., Ltd. | Apparatus and method for user interest information generation |
| CN109376731A (en) * | 2018-08-24 | 2019-02-22 | 北京三快在线科技有限公司 | A kind of character recognition method and device |
| KR102152260B1 (en) | 2020-03-04 | 2020-09-04 | 주식회사 로민 | Apparatus and method for recognizing key-value relationship |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9430704B2 (en) | Image processing system with layout analysis and method of operation thereof | |
| EP2783328B1 (en) | Text detection using multi-layer connected components with histograms | |
| CN111598074B (en) | Edge detection method and device, electronic equipment and storage medium | |
| KR101725501B1 (en) | Method and apparatus for recognizing character | |
| CN108875535B (en) | Image detection method, device and system and storage medium | |
| KR20160014728A (en) | Method and system for recognizing information | |
| JP6642970B2 (en) | Attention area detection device, attention area detection method, and program | |
| CN107545223B (en) | Image recognition method and electronic equipment | |
| JP7026165B2 (en) | Text recognition method and text recognition device, electronic equipment, storage medium | |
| CN108875504B (en) | Image detection method and image detection device based on neural network | |
| JP2014131278A (en) | Method of authenticating printed document | |
| CN110598566A (en) | Image processing method, device, terminal and computer readable storage medium | |
| JP6542230B2 (en) | Method and system for correcting projected distortion | |
| CN112926564A (en) | Picture analysis method, system, computer device and computer-readable storage medium | |
| CN108960247B (en) | Image significance detection method and device and electronic equipment | |
| CN111209908A (en) | Method and device for updating label box, storage medium and computer equipment | |
| CN108875501B (en) | Human body attribute identification method, device, system and storage medium | |
| CN106663317B (en) | Morphological processing method and digital image processing device for digital image | |
| US10970847B2 (en) | Document boundary detection using deep learning model and image processing algorithms | |
| CN111027545A (en) | Card picture mark detection method and device, computer equipment and storage medium | |
| US20230410561A1 (en) | Method and apparatus for distinguishing different configuration states of an object based on an image representation of the object | |
| CN115205113A (en) | Image splicing method, device, equipment and storage medium | |
| CN114550062A (en) | Method and device for determining moving object in image, electronic equipment and storage medium | |
| JP6717769B2 (en) | Information processing device and program | |
| CN113159037A (en) | Picture rectification method and device, computer equipment and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| GRNT | Written decision to grant |