CN115909369A - Method and system for extracting binary slice image of Chinese character font - Google Patents
Method and system for extracting binary slice image of Chinese character font Download PDFInfo
- Publication number
- CN115909369A CN115909369A CN202310116030.8A CN202310116030A CN115909369A CN 115909369 A CN115909369 A CN 115909369A CN 202310116030 A CN202310116030 A CN 202310116030A CN 115909369 A CN115909369 A CN 115909369A
- Authority
- CN
- China
- Prior art keywords
- picture
- handwriting practicing
- grids
- image
- extracting
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
- Y02T10/00—Road transport of goods or passengers
- Y02T10/10—Internal combustion engine [ICE] based vehicles
- Y02T10/40—Engine management systems
Landscapes
- Image Analysis (AREA)
Abstract
The invention discloses a method and a system for extracting a binary slice image of a Chinese character font, which comprises the following steps: processing the acquired RGB handwriting practicing grid calligraphy picture to obtain a gray level picture; extracting a maximum quadrilateral outline from the gray level picture, and shielding a noise background outside the outline; extracting penmanship practice lattice frame lines in the horizontal direction and the vertical direction from the picture with the noise background shielded by adopting an LSD (least squares) linear detection algorithm, and eliminating the influence of dotted line noise of the penmanship practice lattices to obtain a horizontal penmanship practice lattice line marking graph and a vertical penmanship practice lattice line marking graph; determining the actual crossed key points of the handwriting practicing grids to obtain RGB slice images of the handwriting practicing grids; and carrying out cluster analysis on the gray values of the slice images to obtain a binary slice image. The method can avoid the influence of factors such as auxiliary broken lines of the handwriting practicing grids, various smearing and photographing lights and the like, accurately slice the calligraphy pictures of the RGB handwriting practicing grids, and simultaneously binarize the slice pictures to accurately extract the characters of the handwritten Chinese characters.
Description
Technical Field
The invention belongs to the technical field of image processing, and particularly relates to a method and a system for extracting a binary slice image of a Chinese character font.
Background
The Chinese calligraphy is a unique Chinese character art and is an art which can continuously and widely express national spirit and time spirit. However, in the current stage, traditional calligraphy education mainly depends on hand-held teaching of teachers, and extremely depends on teaching modes and teaching resources of the teachers. Therefore, an intelligent evaluation system for evaluating handwritten Chinese characters of students by an intelligent means to give fair and objective suggestions to improve the calligraphy level of the students is particularly important.
The beginners practice calligraphy generally use exercise paper of character-practicing grids, such as rice character grids or field character grids. In the existing calligraphy intelligent evaluation system, uploaded calligraphy pictures need to be sliced and divided into images of single handwriting practicing grids, so that each single character can be evaluated sequentially according to an intelligent scheme. However, in the process of cutting the picture, no matter the auxiliary dotted line in the rice character grid or the field character grid, or the character written by the practicer, even including some factors such as smearing, changing and the like on the paper surface, will bring huge influence and challenge to the accurate segmentation of the picture, and simultaneously, the difference brought by the light and the photographing device when the picture is taken is also considered. However, the existing calligraphy intelligent evaluation system cannot avoid the above influencing factors, and finally, the reasonable and objective suggestions are difficult to be given due to inaccurate calligraphy picture cutting.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a method and a system for extracting a binary slice image of a Chinese character font, which can avoid the influence of factors such as auxiliary broken lines of handwriting practicing grids, various smearing lights, photographing lights and the like, accurately slice an RGB handwriting practicing grid calligraphy picture, and simultaneously binarize a slice image to accurately extract the handwritten Chinese character font.
The invention provides the following technical scheme:
in a first aspect, a method for extracting a binarized sliced image of a Chinese character font is provided, which includes:
acquiring an RGB handwriting practicing grid calligraphy picture, and processing the acquired RGB handwriting practicing grid calligraphy picture to obtain a gray level picture;
extracting a maximum quadrilateral outline from the gray level picture, and shielding the area outside the outline as a noise background to obtain a picture with the noise background shielded;
respectively extracting frame lines of the handwriting practicing grids in horizontal and vertical directions from a picture with a shielded noise background by adopting an LSD (least squares) linear detection algorithm to obtain a horizontal frame line binary image and a vertical frame line binary image;
extracting the edge lines of the handwriting practicing grids by using a Sobel operator to obtain a rough binaryzation handwriting practicing grid outer frame mask map, performing and operation on the horizontal frame line binaryzation map and the vertical frame line binaryzation map with the rough binaryzation handwriting practicing grid outer frame mask map respectively, setting a threshold value, eliminating short line segments in the handwriting practicing grids, and obtaining a horizontal frame line binaryzation optimization map and a vertical frame line binaryzation optimization map;
detecting the proportion of the distance between adjacent lattice lines in the horizontal frame line binarization optimization graph and the vertical frame line binarization optimization graph, and removing redundant lines in the handwriting practicing grids to obtain a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph;
determining actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain RGB slice graphs of the handwriting practicing grids;
and performing K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain a binary slice image of the Chinese character font.
Further, the processing of the acquired RGB handwriting practicing grid images includes:
carrying out mean value filtering on the acquired RGB handwriting practicing grids to obtain a preprocessed picture;
reducing the quality of pre-processed pictures, the rate of quality reductionvComprises the following steps:
v=y w /y n ；
in the formula (I), the compound is shown in the specification,y w representing the width resolution of the pre-processed picture,y n representing the width resolution of the pre-processed picture after the quality is reduced;
enhancing the contrast and the sharpening degree of the preprocessed picture with reduced quality to obtain a pictureSheetg 0 Then to the pictureg 0 Performing graying treatment of the prominent color difference to obtain the pictureg 1 ：
g 1 (i,j)=Max(R(i,j),G(i,j),B(i,j))；
For picturesg 1 Obtaining pictures by Gaussian blurg 2 :
In the formula (I), the compound is shown in the specification, (ii) (i,j) The coordinates of the pixel points of the two-dimensional picture are represented,g 1 (i,j) Representing the coordinates of a pixel point (i,j) The pixel values after contrast and sharpness enhancement,g 2 (i,j) Expressing coordinates of pixel points (i,j) The pixel values after the gaussian blur are obtained,Max(R(i,j),G(i,j),B(i,j) Express picturesg 0 In the point coordinates, the maximum value among the three channel gray values,R(i,j) Presentation pictureg 0 The value of the R-channel at the point coordinate,G(i,j) Presentation pictureg 0 The G-channel value at the point coordinate,B(i,j) Presentation pictureg 0 The B channel value at the point coordinate, ((ii))u,v) The coordinates of the convolution kernel are represented as,rwhich represents the radius of the convolution kernel,s(i+u,j +v) Presentation pictureg 1 At the value of the pixel at that coordinate,fwhich represents a gaussian filter function, is used,f(u,v) Representing a convolution kernel inu,v) The weight value of (1);
for pictureg 2 Performing morphological closed operation to obtain pictureg 3 Then will beg 1 /g 3 Normalizing the result to obtain a gray level picture with outstanding color aberration and reduced light brightness influenceG 0 。
Further, a specific method for extracting the maximum quadrilateral profile from the grayscale image includes: and extracting a line graph of the handwriting practicing grids in the gray level picture by adopting an LSD (least squares) line detection algorithm, and finding out a maximum quadrilateral outline.
Further, the specific method for obtaining the horizontal frame line binarization optimization map and the vertical frame line binarization optimization map comprises the following steps:
calculating picturesg 0 The variance of the gray values of the three channels of each pixel point:
in the formula (I), the compound is shown in the specification,Dx ij expressing coordinates of pixel points (i,j) R, G, B the variance of the gray values of the three channels,R i, j representing the coordinates of a pixel point (i,j) The value of the R-channel of (a),G i, j representing the coordinates of a pixel point (i,j) The value of the G-channel of (c),B i, j representing the coordinates of a pixel point (i,j) The value of the B-channel of (a),mean(x ij ) Representing the coordinates of a pixel point (i,j) R, G, B mean of three channel values;
will be provided withDx ij Normalizing to 0-255 range, simultaneously extracting the edge lines of the handwriting practicing grid by using Sobel operator to obtain rough Mask of the outer frame Mask of the binarization handwriting practicing grid, and binarizing the horizontal frame line binary imageH(x,y) And vertical frame line binary imageV(x,y) Respectively performing AND operation with Mask, simultaneously setting threshold, eliminating short line segments in the handwriting practicing grids, preliminarily removing influence of interference line segments, and obtaining a horizontal frame line binaryzation optimization graphH 0 (x,y) And vertical frame line binaryzation optimization mapV 0 (x,y)；。
Further, the specific method for obtaining the RGB slice images of the handwriting practicing grids comprises:
marking the horizontal handwriting practicing grids with linesHLine(x,y) And vertical handwriting practicing grid line marking chartVLine(x,y) The and operation is performed to carry out the and operation,obtaining a line intersection marking mapdot(x,y)：
Marking a graph according to line intersectionsdot(x,y) Obtaining the centroid of each cross point of the handwriting practicing gridsMoments(x,y) Wherein the center of massMoments(x,y) The calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,m pq representing imagesp+qThe order of the moment is set to be,prepresentxThe number of times of the operation of the motor,qrepresentyThe number of times of the operation of the motor,m 10 to representxThe first order moment of (a) of (b),m 01 representyThe first order moment of (a) of (b),m 00 representing the zeroth order moment of the image;
according to the center of massMoments(x,y) And mass reduction ratiovDetermining actual handwriting practicing grid cross key point coordinates on the acquired RGB handwriting practicing grid calligraphy picture
：
According to the actual handwriting practicing grids, the coordinates of key points are crossed
And the number of rowsnumHNumber of sum columnsnumVIn, to>
Sorting and ordering are carried out, the number of words and the number of columns of each row are determined, and then segmentation is carried out to obtain the RGB slice images of the handwriting practicing grids.
Further, the specific method of the K-means cluster analysis comprises:
calculate allDx ij Pixel point > thresholdT(x,y) The image brightness of three channels R, G, B;
selecting a channel with the maximum picture brightness for K-means clustering, selecting the initial clustering centers as 0, 128 and 255, and selecting the class with the minimum clustering center value from the three clustered classes as the gray value range of the final Chinese character fontRange；
Judging the gray value of the selected channel pictureGrayWhether the gray value range is satisfied;
if the gray value of the selected channel picture
If the pixel point is a certain part of the Chinese character font, the pixel value of the point is set to be 0; otherwise, the pixel value is set to 255, and finally the binary slice image of each single character Chinese character font is obtained.
Further, the pixel pointsT(x,y) Picture luminance in accordance with R, G, B three channelsLight(T R )、Light(T G )、Light(T B ) The calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,R(x,y) The R-channel value representing the coordinates of the picture at that point,G(x,y) A G-channel value representing the coordinates of the picture at that point,B(x,y) A B-channel value representing the coordinates of the picture at that point,length(T(x,y) ) represents a set of pixel pointsCombination of Chinese herbsT(x,y) The number of the pixels in (2).
In a second aspect, a system for extracting a binarized slice image of a chinese character font is provided, which includes:
the data acquisition and processing module is used for acquiring the RGB handwriting practicing grids and processing the acquired RGB handwriting practicing grids to obtain a gray level picture;
the first extraction module is used for extracting a maximum quadrilateral outline from the gray level picture and shielding an area outside the outline as a noise background to obtain a picture with the noise background shielded;
the second extraction module is used for extracting the frame lines of the handwriting practicing grids in the horizontal direction and the vertical direction respectively in the picture with the noise background shielded by adopting an LSD (least squares) linear detection algorithm to obtain a horizontal frame line binary image and a vertical frame line binary image;
the first image processing module is used for extracting the edge lines of the handwriting practicing grids by using a Sobel operator to obtain a rough binaryzation handwriting practicing grid outer frame mask image, performing AND operation on the horizontal frame line binaryzation image and the vertical frame line binaryzation image with the rough binaryzation handwriting practicing grid outer frame mask image respectively, setting a threshold value, eliminating short line segments in the handwriting practicing grids, and obtaining a horizontal frame line binaryzation optimization image and a vertical frame line binaryzation optimization image;
the second picture processing module is used for detecting the proportion of the distance between adjacent lattice lines in the horizontal frame line binarization optimization graph and the vertical frame line binarization optimization graph, eliminating redundant lines in the handwriting practicing grids and obtaining a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph;
the third picture processing module is used for determining the actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain RGB slice graphs of the handwriting practicing grids;
and the cluster analysis module is used for performing K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain the binaryzation slice images of the Chinese character fonts.
In a third aspect, an apparatus for extracting a binarized slice image of a Chinese character font is provided, which includes a processor and a storage medium; the storage medium is used for storing instructions; the processor is configured to operate in accordance with the instructions to perform the steps of the method of the first aspect.
In a fourth aspect, a computer-readable storage medium is provided, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to the first aspect.
Compared with the prior art, the invention has the beneficial effects that:
(1) The maximum quadrilateral outline is extracted from the gray level picture, the area outside the outline is used as the noise background to be shielded, the picture with the noise background shielded is obtained, the LSD linear detection algorithm is adopted to respectively extract the frame lines of the handwriting practicing grids in the horizontal and vertical directions from the picture with the noise background shielded, the influence of the dotted line noise of the handwriting practicing grids is eliminated, and the horizontal handwriting practicing grids line marking graph and the vertical handwriting practicing grids line marking graph are obtained, so that the influence of factors such as auxiliary dotted lines of the handwriting practicing grids, various smearing rays, photographing rays and the like is avoided, the actual crossed key points of the handwriting practicing grids are more accurately found, and the accurate RGB slice graph of each handwriting practicing grid is obtained;
(2) The method eliminates the influence of the auxiliary dotted lines and the outer frame of the handwriting practicing grids through the K-means clustering analysis, accurately extracts the handwritten Chinese character fonts, lays a foundation for the calligraphy intelligent evaluation system to subsequently extract the Chinese character frameworks, perform analysis evaluation according to the stroke position and shape and give improvement suggestions, and meets the precondition requirements of the calligraphy evaluation system.
Drawings
FIG. 1 is a flow chart of a method for extracting a binarized sliced image of Chinese character fonts according to an embodiment of the present invention;
FIG. 2 is a diagram of an original RGB field grid calligraphy picture obtained in step 1 according to an embodiment of the present invention;
FIG. 3 is a grayscale image outputted in step 2 according to the embodiment of the present invention;
FIG. 4 is a picture of the masked noise background output in step 3 according to the embodiment of the present invention;
FIG. 5 is a horizontal grid line marking graph (a) and a vertical grid line marking graph (b) output in step 4 according to the embodiment of the present invention;
FIG. 6 is an RGB slice diagram of two grid fields output in step 5 according to an embodiment of the present invention;
fig. 7 is a binarized slice image of two chinese character fonts output in step 6 according to an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the accompanying drawings. The following examples are only for illustrating the technical solutions of the present invention more clearly, and the protection scope of the present invention is not limited thereby.
Example 1
As shown in fig. 1, the present embodiment provides a method for extracting a binarized sliced image of a chinese character font, including the following steps:
step 1, acquiring an RGB handwriting practicing grid picture shot by a user.
And 2, processing the RGB handwriting practicing grid pictures obtained in the step 1 to obtain gray pictures, highlighting color difference and reducing the influence of light brightness. The method comprises the following specific steps:
(2.1) carrying out mean value filtering on the RGB handwriting practicing grids to obtain a preprocessed picture after noise reduction, wherein the adopted filtering operator is shown in a table 1.
TABLE 1
| 1/9 | 1/9 | 1/9 |
| 1/9 | 1/9 | 1/9 |
| 1/9 | 1/9 | 1/9 |
(2.2) reducing the quality of the preprocessed picture, specifically, reducing the fixed-width pixels of the preprocessed picture, adjusting the height adaptively, and reducing the quality reduction ratiovComprises the following steps:
v=y w /y n ；
in the formula (I), the compound is shown in the specification,y w representing the width resolution of the pre-processed picture,y n indicating the width resolution of the degraded pre-processed picture.
(2.3) enhancing the contrast and the sharpening degree of the low-quality preprocessed picture to obtain the pictureg 0 Then to the pictureg 0 Performing graying treatment of the prominent color difference to obtain a grayed pictureg 1 ：
g 1 (i,j)=Max(R(i,j),G(i,j),B(i,j))；
For picturesg 1 Obtaining pictures by Gaussian blurg 2 :
In the formula (I), the compound is shown in the specification, (ii) (i,j) The coordinates of the pixels of the two-dimensional picture are represented,g 1 (i,j) Representing the coordinates of a pixel point (i,j) The pixel values after contrast and sharpening enhancement,g 2 (i,j) Representing the coordinates of a pixel point (i,j) The pixel values after the gaussian blur are obtained,Max(R(i,j),G(i,j),B(i,j) Express picturesg 0 In the point coordinates, the maximum value among the three channel gray values,R(i,j) Presentation pictureg 0 The value of the R-channel at the point coordinate,G(i,j) Presentation pictureg 0 The value of the R-channel at the point coordinate,B(i,j) Presentation pictureg 0 The R channel value at the point coordinate(s) ((s))u,v) The coordinates of the convolution kernel are represented as,rwhich represents the radius of the convolution kernel,s(i+u,j +v) Presentation pictureg 1 At the value of the pixel at that coordinate,fwhich represents a gaussian filter function, is used,f(u,v) Representing convolution kernels in (u,v) The weight value of (3).
(2.4) for picturesg 2 Performing morphological closed operation to obtain pictureg 3 Then will beg 1 /g 3 Normalizing the result to obtain a gray level picture with outstanding color aberration and reduced light brightness influenceG 0 。
And 3, extracting the maximum quadrilateral outline from the gray level picture, and shielding the area outside the outline as a noise background to obtain the picture with the noise background shielded. The method comprises the following specific steps:
(3.1) extracting a handwriting practicing grid line graph in the gray level picture by adopting an LSD (least squares) line detection algorithm, simultaneously performing expansion operation in morphology, connecting similar line segments, then finding a contour with the largest area, recording all vertex coordinates of the contour as N, executing a convex hull algorithm Graham scan, and finding a contour which is largest and approximate to a quadrangle.
The logic idea of the convex hull algorithm Graham scan is as follows: solving a point p with the minimum y value in the N points; sorting the rest N-1 points according to the polar angle value of the point p; traversing the sorted N-1 points, and only keeping the points rotating anticlockwise.
And (3.2) shielding the region outside the outline as a noise background, and simultaneously performing perspective correction to obtain an input picture of the next step.
And 4, extracting the frame lines of the handwriting practicing grids in the horizontal direction and the vertical direction from the picture with the noise background shielded, and eliminating the influence of the dotted line noise of the handwriting practicing grids to obtain a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph. The method comprises the following specific steps:
(4.1) adopting an LSD linear detection algorithm to respectively extract the frame lines of the handwriting practicing grids in the horizontal and vertical directions from the picture with the shielded noise background to obtain a binarization graph of the horizontal frame linesH(x,y) And a vertical frame line binary imageV(x,y) Because of the influence of a large number of interference auxiliary dotted lines and strokes of Chinese characters in the handwriting practicing grids, the horizontal frame line binary image at the momentH(x,y) And a vertical frame line binary imageV(x,y) Some redundant lines and interference line segments may appear in the image, and need to be further removed.
(4.2) calculating the Pictureg 0 The variance of the gray values of the three channels of each pixel point:
in the formula (I), the compound is shown in the specification,Dx ij representing the coordinates of a pixel point (i,j) R, G, B the variance of the gray values of the three channels,R i, j expressing coordinates of pixel points (i,j) The value of the R-channel of (a),G i, j expressing coordinates of pixel points (i,j) The value of the G-channel of (a),B i, j expressing coordinates of pixel points (i,j) The value of the B-channel of (a),mean(x ij ) Representing the coordinates of a pixel point (i,j) R, G, B.
(4.3) mixingDx ij Normalizing to 0-255 range, simultaneously extracting the edge lines of the handwriting practicing grid by using Sobel operator to obtain rough Mask of the outer frame Mask of the binarization handwriting practicing grid, and binarizing the horizontal frame line binary imageH(x,y) And vertical frame line binary imageV(x,y) Respectively performing AND operation with Mask, simultaneously setting threshold, eliminating short line segments in the handwriting practicing grids, preliminarily removing influence of interference line segments, and obtaining a horizontal frame line binaryzation optimization graphH 0 (x,y) And vertical frame line binaryzation optimization mapV 0 (x,y)。
(4.4) detecting the horizontal frame line binaryzation optimization mapH 0 (x,y) And vertical frame line binaryzation optimization mapV 0 (x,y) The ratio of the distance between adjacent grid lines. Let us note thatnThe distance between adjacent lattice lines is Dist: (n) Of 1 atn+1 distance between adjacent lattice lines Dist ()n+ 1), the ratio of the distance between two adjacent grid lines is:
ratio(n)=Dist(n)/ Dist(n+1)；
judging whether the k-th line is a redundant line, if so, removing the line, otherwise, continuously judging:
signal = 1
while signal:
for k in lines:
ifratio(k-1)/ratio(k)>1.42 and ratio(k)/ratio(k+1)<0.7:
drop the line of k
continue
signal = 0
finally, all redundant lines in the handwriting practicing grids are removed to obtain a marking graph of the horizontal handwriting practicing grids linesHLine(x,y) And vertical handwriting practicing grid line marking pictureVLine(x,y) Simultaneously, the number of rows and columns is obtained according to the number of lines of each graphnumHAndnumV。
and 5, determining actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain the RGB slice graph of each handwriting practicing grid. The method comprises the following specific steps:
(5.1) marking the horizontal handwriting practicing grids with linesHLine(x,y) And vertical handwriting practicing grid line marking chartVLine(x,y) Performing AND operation to obtain a line intersection marking mapdot(x,y)：
(5.2) marking the graph according to the line intersection pointsdot(x,y) Obtaining the centroid of each cross point of the handwriting practicing gridsMoments(x,y) Wherein the center of massMoments(x,y) The calculation formula of (c) is:
in the formula (I), the compound is shown in the specification,m pq representing imagesp+qThe order of the moment of the wave,prepresentxThe number of times of the operation of the motor,qto representyThe number of times of the above-mentioned operations,m 10 representxThe first order moment of (a) of (b),m 01 representyThe first moment of (a) is,m 00 representing the zeroth order moment of the image.
(5.3) according to the centroidMoments(x,y) And mass reduction ratiovDetermining the actual handwriting practicing grid cross key point coordinates on the RGB handwriting practicing grid calligraphy picture acquired in the step 1
：
According to the actual handwriting practicing grids, the coordinates of key points are crossed
And the number of rowsnumHNumber of sum columnsnumVIn, to>
Sorting and ordering are carried out, the number of words and the number of columns of each row are determined, and then segmentation is carried out to obtain the RGB slice images of the handwriting practicing grids.
And 6, performing K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain the binary slice images of the Chinese character fonts. The method comprises the following specific steps:
(6.1) variance ofDx ij If the variance threshold is greater than the threshold, the color of the grid of the handwriting practicing grids in the picture is prominent, and all the statistics is carried outDx ij Pixel point greater than thresholdT(x,y) Calculating all pixel pointsT(x,y) Picture brightness according to R, G, B three channelsLight(T R )、Light(T G )、Light(T B )：
In the formula (I), the compound is shown in the specification,R(x,y) The R-channel value representing the coordinates of the picture at that point,G(x,y) A G-channel value representing the coordinates of the picture at that point,B(x,y) A B-channel value representing the coordinates of the picture at that point,length(T(x,y) ) represents a collection of pixel pointsT(x,y) The number of the pixels in (1).
(6.2) selecting a channel with the maximum picture brightness for K-means clustering, selecting initial clustering centers as 0, 128 and 255, namely a white background class, a handwriting practicing grid line class and a calligraphy font class, and selecting the class with the minimum clustering center value in the three clustered classes as the gray value range of the final Chinese character fontRange。
(6.3) judging the gray value of the selected channel pictureGrayWhether the gray value range is satisfied; if the gray value of the selected channel picture
If the pixel point is a certain part of the Chinese character font, the pixel value of the point is set to be 0; otherwise, the pixel value is set to 255, and finally the binary slice image of each single character Chinese character font is obtainedB(x,y)。
Example 2
The method in embodiment 1 is adopted to extract the binary slice image of the Chinese character font from the RGB field character grid calligraphy picture.
FIG. 2 is a diagram of an acquired original RGB field grid calligraphy picture. The fixed width pixels after the pre-processed picture is reduced in step 2.2 to 800 pixels, i.e.v=y w And 800, and fig. 3 is the gray picture output in the step 2. Fig. 4 is a picture of the masked noise background output through step 3. Fig. 5 (a) and 5 (b) are a horizontal grid line marking diagram and a vertical grid line marking diagram output after step 4, respectively. Fig. 6 is an RGB slice image of two mattes obtained after step 5. And FIG. 7 is the binarized slice image of the Chinese character font output in step 6, which shows that the binarized slice image more completely embodies the characteristics and details of the Chinese character written by the user, and lays a foundation for the calligraphy intelligent evaluation system to subsequently extract the Chinese character skeleton, perform analysis and evaluation according to the stroke position and shape, and give improvement suggestions.
Example 3
The embodiment provides an extracting system of a binarized sliced image of a Chinese character font, which comprises:
the data acquisition and processing module is used for acquiring the RGB handwriting practicing grids and processing the acquired RGB handwriting practicing grids to obtain a gray level picture;
the first extraction module is used for extracting a maximum quadrilateral outline from the gray level picture and shielding an area outside the outline as a noise background to obtain a picture with the noise background shielded;
the second extraction module is used for respectively extracting the frame lines of the handwriting practicing grids in the horizontal direction and the vertical direction from the picture with the noise background shielded by adopting an LSD (least squares) linear detection algorithm to obtain a horizontal frame line binary image and a vertical frame line binary image;
the first image processing module is used for extracting the edge lines of the handwriting practicing grids by using a Sobel operator to obtain a rough binaryzation handwriting practicing grid outer frame mask image, performing AND operation on the horizontal frame line binaryzation image and the vertical frame line binaryzation image with the rough binaryzation handwriting practicing grid outer frame mask image respectively, setting a threshold value, eliminating short line segments in the handwriting practicing grids, and obtaining a horizontal frame line binaryzation optimization image and a vertical frame line binaryzation optimization image;
the second picture processing module is used for detecting the proportion of the distance between adjacent lattice lines in the horizontal frame line binarization optimization graph and the vertical frame line binarization optimization graph, eliminating redundant lines in the handwriting practicing grids and obtaining a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph;
the third picture processing module is used for determining the actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain RGB slice graphs of the handwriting practicing grids;
and the cluster analysis module is used for carrying out K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain the binary slice images of the Chinese character fonts.
Example 4
The embodiment provides an extraction device of a binaryzation slice image of a Chinese character font, which comprises a processor and a storage medium; the storage medium is to store instructions; the processor is configured to operate in accordance with the instructions to perform the steps of the method of embodiment 1.
Example 5
The present embodiment provides a computer-readable storage medium having stored thereon a computer program which, when executed by a processor, performs the steps of the method of embodiment 1.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
The above description is only a preferred embodiment of the present invention, and it should be noted that, for those skilled in the art, several modifications and variations can be made without departing from the technical principle of the present invention, and these modifications and variations should also be regarded as the protection scope of the present invention.
Claims (10)
1. A method for extracting a binary slice image of a Chinese character font is characterized by comprising the following steps:
acquiring an RGB handwriting practicing grid calligraphy picture, and processing the acquired RGB handwriting practicing grid calligraphy picture to obtain a gray level picture;
extracting a maximum quadrilateral outline from the gray level picture, and shielding the area outside the outline as a noise background to obtain a picture with the noise background shielded;
respectively extracting frame lines of the handwriting practicing grids in horizontal and vertical directions from a picture with a shielded noise background by adopting an LSD (least squares) linear detection algorithm to obtain a horizontal frame line binary image and a vertical frame line binary image;
extracting the edge lines of the handwriting practicing grids by using a Sobel operator to obtain a rough binaryzation handwriting practicing grid outer frame mask map, performing and operation on the horizontal frame line binaryzation map and the vertical frame line binaryzation map with the rough binaryzation handwriting practicing grid outer frame mask map respectively, setting a threshold value, eliminating short line segments in the handwriting practicing grids, and obtaining a horizontal frame line binaryzation optimization map and a vertical frame line binaryzation optimization map;
detecting the proportion of the distance between adjacent grid lines in the horizontal frame line binarization optimization graph and the vertical frame line binarization optimization graph, and removing redundant lines in the handwriting practicing grids to obtain a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph;
determining actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain RGB slice graphs of the handwriting practicing grids;
and performing K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain a binary slice image of the Chinese character font.
2. The method for extracting binarized sliced image from Chinese character fonts according to claim 1, wherein the processing of the acquired RGB handwriting practicing grids comprises:
carrying out mean value filtering on the acquired RGB handwriting practicing grids to obtain a preprocessed picture;
reducing the quality of pre-processed pictures, the rate of quality reductionvComprises the following steps:
v=y w /y n ；
in the formula (I), the compound is shown in the specification,y w representing the width resolution of the pre-processed picture,y n representing the width resolution of the pre-processed picture after the quality is reduced;
enhancing the contrast and the sharpening degree of the pre-processed picture with reduced quality to obtain a pictureg 0 Then to the pictureg 0 Performing graying treatment of the prominent color difference to obtain the pictureg 1 ：
g 1 (i, j)= Max(R(i,j),G(i,j),B(i,j))；
For pictureg 1 Obtaining pictures by Gaussian blurg 2 :
Wherein (A) isi, j) The coordinates of the pixel points of the two-dimensional picture are represented,g 1 (i, j) Expressing coordinates of pixel points (i, j) The pixel values after contrast and sharpness enhancement,g 2 (i, j) Representing the coordinates of a pixel point (i, j) The pixel values after the gaussian blur are obtained,Max(R(i,j),G(i,j),B(i,j) Express picturesg 0 In the point coordinates, the maximum value among the three channel gray values,R(i,j) Representing picturesg 0 The value of the R-channel at the point coordinate,G(i,j) Presentation pictureg 0 The G-channel value at the point coordinate,B(i,j) Presentation pictureg 0 The B channel value at the point coordinate, ((ii))u, v) The coordinates of the convolution kernel are represented as,rwhich represents the radius of the convolution kernel,s(i+ u,j + v) Representing picturesg 1 At the value of the pixel at that coordinate,fwhich represents a gaussian filter function, is used,f(u, v) Representing a convolution kernel inu, v) OnA weight value;
for picturesg 2 Performing morphological closed operation to obtain pictureg 3 Then will beg 1 /g 3 Normalizing the result to obtain a gray level picture with outstanding color aberration and reduced light brightness influenceG 0 。
3. The method for extracting the binarized sliced image for Chinese character fonts according to claim 1, wherein the specific method for extracting the maximum outline of the quadrangle from the gray-scale picture comprises the following steps: and extracting a line graph of the handwriting practicing grids in the gray level picture by adopting an LSD (least squares) line detection algorithm, and finding out a maximum quadrilateral outline.
4. The method for extracting the binarized sliced image of Chinese character fonts according to claim 2, wherein the specific method for obtaining the binarized optimized graph of horizontal border lines and the binarized optimized graph of vertical border lines comprises the following steps:
calculating picturesg 0 The variance of the gray values of the three channels of each pixel point:
in the formula (I), the compound is shown in the specification,Dx ij expressing coordinates of pixel points (i, j) R, G, B the variance of the gray values of the three channels,R i, j expressing coordinates of pixel points (i, j) The value of the R-channel of (a),G i, j representing the coordinates of a pixel point (i, j) The value of the G-channel of (a),B i, j expressing coordinates of pixel points (i, j) The value of the B-channel of (a),mean(x ij ) Expressing coordinates of pixel points (i, j) R, G, B mean of three channel values;
will be provided withDx ij Normalizing to 0-255 range, simultaneously extracting the edge lines of the handwriting practicing grid by using Sobel operator to obtain rough Mask of the outer frame Mask of the binarization handwriting practicing grid, and binarizing the horizontal frame line binary imageH(x,y) And vertical frame line binary imageV(x, y) Respectively performing AND operation with Mask, setting threshold value simultaneously, eliminating short line segment in handwriting practicing grid, preliminarily removing influence of interference line segment, and obtaining horizontal frame line binaryzation optimization mapH 0 (x,y) And vertical frame line binaryzation optimization graphV 0 (x, y)。
5. The method for extracting binarized slice image for Chinese character fonts according to claim 1, wherein the specific method for obtaining RGB slice images of handwriting practicing grids comprises the following steps:
marking the horizontal handwriting practicing grids with linesHLine(x,y) And vertical handwriting practicing grid line marking chartVLine (x, y) Performing AND operation to obtain a line intersection marking mapdot(x,y)：
Marking a graph according to line intersectionsdot(x,y) Obtaining the centroid of each cross point of the handwriting practicing gridsMoments(x,y) Wherein the center of massMoments(x,y) The calculation formula of (2) is as follows:
in the formula (I), the compound is shown in the specification,m pq representing imagesp+qThe order of the moment is set to be,pto representxThe number of times of the above-mentioned operations,qto representyThe number of times of the above-mentioned operations,m 10 to representxThe first moment of (a) is,m 01 to representyThe first moment of (a) is,m 00 representing the zeroth order moment of the image;
according to the center of massMoments(x,y) And massDecrease the ratiovDetermining actual handwriting practicing grid cross key point coordinates on the acquired RGB handwriting practicing grid calligraphy picture
：
According to the actual handwriting practicing grids, the coordinates of key points are crossed
And number of rowsnumHNumber of sum columnsnumVIn, to>
Sorting and sequencing are carried out, the number of words and the number of columns of each row are determined, and then segmentation is carried out to obtain RGB slice images of all handwriting practicing grids.
6. The method for extracting the binarized sliced image of Chinese character fonts according to claim 4, wherein the specific method of the K-means cluster analysis comprises the following steps:
calculate allDx ij Pixel point > thresholdT(x,y) The image brightness of three channels R, G, B;
selecting a channel with the maximum picture brightness for K-means clustering, selecting the initial clustering centers as 0, 128 and 255, and selecting the class with the minimum clustering center value from the three clustered classes as the gray value range of the final Chinese character fontRange；
Judging the gray value of the selected channel pictureGrayWhether the gray value range is satisfied;
if the gray value of the selected channel picture
If the pixel point is a certain part of the Chinese character font, the pixel value of the point is set to be 0; otherwise, the pixel value is set to 255, and finally binaryzation of each single character Chinese character font is obtainedAnd (4) slicing the image.
7. The method for extracting binarized sliced image for Chinese character fonts as recited in claim 6, wherein said pixels are located in a same plane as said reference planeT(x,y) Picture luminance in accordance with R, G, B three channelsLight(T R )、Light(T G )、Light(T B ) The calculation formula of (c) is:
in the formula (I), the compound is shown in the specification,R(x,y) The R-channel value representing the coordinates of the picture at that point,G(x,y) A G-channel value representing the coordinates of the picture at that point,B(x,y) A B-channel value representing the coordinates of the picture at that point,length(T(x,y) ) represents a collection of pixel pointsT(x,y) The number of the pixels in (1).
8. A Chinese character font binarization slice image extraction system is characterized by comprising:
the data acquisition and processing module is used for acquiring the RGB handwriting practicing grids and processing the acquired RGB handwriting practicing grids to obtain a gray level picture;
the first extraction module is used for extracting a maximum quadrilateral outline from the gray level picture and shielding an area outside the outline as a noise background to obtain a picture with the noise background shielded;
the second extraction module is used for respectively extracting the frame lines of the handwriting practicing grids in the horizontal direction and the vertical direction from the picture with the noise background shielded by adopting an LSD (least squares) linear detection algorithm to obtain a horizontal frame line binary image and a vertical frame line binary image;
the first image processing module is used for extracting the edge lines of the handwriting practicing grids by using a Sobel operator to obtain a rough binaryzation handwriting practicing grid outer frame mask image, performing AND operation on the horizontal frame line binaryzation image and the vertical frame line binaryzation image with the rough binaryzation handwriting practicing grid outer frame mask image respectively, setting a threshold value, eliminating short line segments in the handwriting practicing grids, and obtaining a horizontal frame line binaryzation optimization image and a vertical frame line binaryzation optimization image;
the second picture processing module is used for detecting the proportion of the distance between adjacent lattice lines in the horizontal frame line binarization optimization graph and the vertical frame line binarization optimization graph, eliminating redundant lines in the handwriting practicing grids and obtaining a horizontal handwriting practicing grid line marking graph and a vertical handwriting practicing grid line marking graph;
the third picture processing module is used for determining the actual crossed key points of the handwriting practicing grids according to the horizontal handwriting practicing grid line marking graph and the vertical handwriting practicing grid line marking graph to obtain RGB slice graphs of the handwriting practicing grids;
and the cluster analysis module is used for performing K-means cluster analysis on the gray values of the RGB slice images of the handwriting practicing grids to obtain the binaryzation slice images of the Chinese character fonts.
9. An extraction device of a Chinese character font binarization slice image is characterized by comprising a processor and a storage medium; the storage medium is to store instructions; the processor is configured to operate in accordance with the instructions to perform the steps of the method of any of claims 1~7.
10. A computer readable storage medium having stored thereon a computer program, the program when executed by a processor implementing the steps of the method of any of claims 1~7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310116030.8A CN115909369B (en) | 2023-02-15 | 2023-02-15 | Method and system for extracting binary slice image of Chinese character font |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310116030.8A CN115909369B (en) | 2023-02-15 | 2023-02-15 | Method and system for extracting binary slice image of Chinese character font |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115909369A true CN115909369A (en) | 2023-04-04 |
| CN115909369B CN115909369B (en) | 2023-06-06 |
Family
ID=86471290
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310116030.8A Active CN115909369B (en) | 2023-02-15 | 2023-02-15 | Method and system for extracting binary slice image of Chinese character font |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115909369B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108805076A (en) * | 2018-06-07 | 2018-11-13 | 浙江大学 | The extracting method and system of environmental impact assessment report table word |
| US10169650B1 (en) * | 2017-06-30 | 2019-01-01 | Konica Minolta Laboratory U.S.A., Inc. | Identification of emphasized text in electronic documents |
| CN110647795A (en) * | 2019-07-30 | 2020-01-03 | 正和智能网络科技(广州)有限公司 | Form recognition method |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN112634262A (en) * | 2020-12-31 | 2021-04-09 | 浙江优学智能科技有限公司 | Writing quality evaluation method based on Internet |
| CN113537227A (en) * | 2021-06-28 | 2021-10-22 | 杭州电子科技大学 | Structured text recognition method and system |
| CN114004204A (en) * | 2021-10-28 | 2022-02-01 | 上海交通大学 | Table structure reconstruction and character extraction method and system based on computer vision |
| CN115661848A (en) * | 2022-07-11 | 2023-01-31 | 上海通办信息服务有限公司 | Form extraction and identification method and system based on deep learning |
-
2023
- 2023-02-15 CN CN202310116030.8A patent/CN115909369B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10169650B1 (en) * | 2017-06-30 | 2019-01-01 | Konica Minolta Laboratory U.S.A., Inc. | Identification of emphasized text in electronic documents |
| CN108805076A (en) * | 2018-06-07 | 2018-11-13 | 浙江大学 | The extracting method and system of environmental impact assessment report table word |
| CN110647795A (en) * | 2019-07-30 | 2020-01-03 | 正和智能网络科技(广州)有限公司 | Form recognition method |
| CN111027546A (en) * | 2019-12-05 | 2020-04-17 | 北京嘉楠捷思信息技术有限公司 | Character segmentation method and device and computer readable storage medium |
| CN112634262A (en) * | 2020-12-31 | 2021-04-09 | 浙江优学智能科技有限公司 | Writing quality evaluation method based on Internet |
| CN113537227A (en) * | 2021-06-28 | 2021-10-22 | 杭州电子科技大学 | Structured text recognition method and system |
| CN114004204A (en) * | 2021-10-28 | 2022-02-01 | 上海交通大学 | Table structure reconstruction and character extraction method and system based on computer vision |
| CN115661848A (en) * | 2022-07-11 | 2023-01-31 | 上海通办信息服务有限公司 | Form extraction and identification method and system based on deep learning |
Non-Patent Citations (1)
| Title |
|---|
| 谢亮: "表格识别预处理技术与表格字符提取算法的研究", 信息科技, pages 2 - 4 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115909369B (en) | 2023-06-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108596166A (en) | A kind of container number identification method based on convolutional neural networks classification | |
| CN104751187B (en) | Meter reading automatic distinguishing method for image | |
| CN111310760B (en) | Method for detecting alpha bone inscription characters by combining local priori features and depth convolution features | |
| CN107784669A (en) | A kind of method that hot spot extraction and its barycenter determine | |
| Alkoffash et al. | A survey of digital image processing techniques in character recognition | |
| CN107122777A (en) | A kind of vehicle analysis system and analysis method based on video file | |
| CN104112132A (en) | Automatic gun number identification method | |
| CN108133216B (en) | Nixie tube reading identification method capable of realizing decimal point reading based on machine vision | |
| CN105447834B (en) | A kind of mahjong image irradiation inequality bearing calibration of feature based classification | |
| CN112712273B (en) | Handwriting Chinese character aesthetic degree judging method based on skeleton similarity | |
| CN110598566A (en) | Image processing method, device, terminal and computer readable storage medium | |
| CN110443235B (en) | Intelligent paper test paper total score identification method and system | |
| CN111754538B (en) | Threshold segmentation method for USB surface defect detection | |
| CN110674812B (en) | Civil license plate positioning and character segmentation method facing complex background | |
| CN113705673B (en) | Text detection method, text detection device, text detection equipment and storage medium | |
| CN113158977B (en) | Image character editing method for improving FANnet generation network | |
| CN112446262A (en) | Text analysis method, text analysis device, text analysis terminal and computer-readable storage medium | |
| CN105404868A (en) | Interaction platform based method for rapidly detecting text in complex background | |
| CN115063785B (en) | Method and device for positioning license plate in expressway scene by using target recognition model | |
| CN114926407A (en) | Steel surface defect detection system based on deep learning | |
| CN109741273A (en) | A kind of mobile phone photograph low-quality images automatically process and methods of marking | |
| CN111062331A (en) | Mosaic detection method and device for image, electronic equipment and storage medium | |
| CN112446259A (en) | Image processing method, device, terminal and computer readable storage medium | |
| CN112686265A (en) | Hierarchic contour extraction-based pictograph segmentation method | |
| CN116824608A (en) | Answer sheet layout analysis method based on target detection technology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |