CN104992176B - A kind of Chinese character extracting method towards an inscription on a tablet - Google Patents
A kind of Chinese character extracting method towards an inscription on a tablet Download PDFInfo
- Publication number
- CN104992176B CN104992176B CN201510441259.4A CN201510441259A CN104992176B CN 104992176 B CN104992176 B CN 104992176B CN 201510441259 A CN201510441259 A CN 201510441259A CN 104992176 B CN104992176 B CN 104992176B
- Authority
- CN
- China
- Prior art keywords
- chinese character
- connected domain
- image
- inscription
- tablet
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/20—Image preprocessing
- G06V10/26—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion
- G06V10/267—Segmentation of patterns in the image field; Cutting or merging of image elements to establish the pattern region, e.g. clustering-based techniques; Detection of occlusion by performing operations on regions, e.g. growing, shrinking or watersheds
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
- G06V30/28—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet
- G06V30/287—Character recognition specially adapted to the type of the alphabet, e.g. Latin alphabet of Kanji, Hiragana or Katakana characters
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Character Input (AREA)
- Image Analysis (AREA)
Abstract
The present invention is a kind of Chinese character extracting method towards an inscription on a tablet, mainly includes the Chinese character area positioning method based on Area Voronoi and two parts of individual character cutting method based on connected domain and bounding box.First against the regularity of distribution and feature of Chinese character in an inscription on a tablet image, the connected domain in image is obtained as generation member, and the profile of each connected domain is represented using Freeman coding strands.The Area Voronoi diagrams of image are constructed using the method on extended contour border.Then according to the standard deviation between connected domain, adjacent Area Voronoi areas are merged, so as to complete the positioning action in Chinese character region.After Chinese character area image is obtained, it is pre-processed, removes certain influence of noise.By concluding the design feature of Chinese character, connected domain analysis method and bounding box are overlapped into both determining methods and are combined, character extraction is carried out to Chinese character.The present invention is directed to the arrangement regulation of ancient times an inscription on a tablet, can preferably complete the extraction work of individual Chinese character, and have certain noise immunity.
Description
Technical field
The invention belongs to computer vision and computer graphic image process field, specifically a kind of image Chinese version
Positioning and the method for character extraction, this method can facilitate the research for ancient times Chinese character, to probing into Chinese character development law tool at all times
It is of great significance.
Background technology
An inscription on a tablet, as a kind of traditional art form of calligraphy, it have recorded Chinese character development from ancient times to the present and differentiation
Course.But because an inscription on a tablet typically all lives through the very long history age, while the influence of spontaneous corrosion etc. is further suffered from, make it not
Beneficial to permanent preservation and propagation.With the popularization of computer and the development of related scans equipment, an inscription on a tablet can be scanned into
Digital picture stores in a computer, effectively extracts the Chinese character in an inscription on a tablet to follow-up structure analysis method and probes into Gu
The development law tool of modern Chinese character is of great significance.
The method for carrying out String localization to image at present is main following three kinds:First, the method based on connected domain, such as Zhang
Et al. propose a framework based on high-order markov random file, can be according to maximum-likelihood criterion and node state after
Test whether probabilistic determination is Chinese character region.Such algorithm has good locating effect for the more single image of background, but
It is that its performance generates dependent on initial connected domain, then faces test for the complex image of background texture;Second, it is based on side
The method of edge, the high-contrast that such method is laid particular emphasis between word and background, time efficiency is higher, but can not remove complexity
Grain background, often it is combined with other algorithms, as initial coarse positioning;Third, the method, Wu et al. based on texture utilizes three
The Gaussian filter of kind different scale is filtered to image and carries out linear transformation, it is proposed that a kind of multi-dimension texture segmentation
Text positioning method.Such method generally has higher robustness, but due to needing to carry out the computationally intensive, multiple of texture analysis
Miscellaneous degree is high, more time-consuming.
Research for the progress Chinese character extraction of an inscription on a tablet image at this stage is simultaneously few, because the Chinese character arrangement in an inscription on a tablet is not united
One, the structure of Chinese character is complex so that in general Chinese character extracting method can not be directly applied for the field so as to obtain preferably
Effect.
The content of the invention
To overcome disadvantages mentioned above, it is an object of the invention to probe into the Chinese character Regional Distributing Regularity in an inscription on a tablet image, research
The design feature of Chinese character, while existing String localization and individual character segmentation technique is analyzed, it is that the Chinese character of an inscription on a tablet image obtains proposition
A kind of simple and effective solution.
In order to achieve the above object, the present invention proposes a kind of Chinese character extracting method towards an inscription on a tablet, it is characterised in that bag
Include following steps:
Step (1), using extended boundary method build an inscription on a tablet image Area Voronoi diagrams, each Voronoi areas
One connected domain of domain representation, whether each connected domain is according to wherein comprising Chinese character being divided into Chinese character connected domain and connected with non-Chinese character
Domain;Due to the discreteness of Chinese character, the separation between individual Chinese character stroke can be caused during Area Voronoi diagrams are built,
So as to belong in different connected domains, therefore the component of part Chinese character, referred to as Chinese character is only included for Chinese character connected domain;
Step (2), according to the otherness between connected domain element width standard deviation, go unless Chinese character connected domain, and merge
All Chinese character connected domains, so as to orient the whole Chinese character region in an inscription on a tablet image;
Step (3), the certain noise spot of TV smooth operations removal is carried out to Chinese character area image, carries out binaryzation afterwards,
So that image only facilitates subsequent operation comprising two kinds of pixels of black and white;The row projection histogram in bianry image is calculated, according to an inscription on a tablet
The intermittent rule of calligraphy column distribution, it is partitioned into single-row Chinese character region;
Connected domain in step (4), extraction Chinese character image row, the connected domain may include single complete Chinese character, also may be used
Can be the combination of the different piece of different Chinese character, it is therefore desirable to which connected domain is recombinated;Analyze nine kinds of common Hanzi structures
Feature, the relation between connected domain boundary rectangle is summarized as three kinds of situations:Upper and lower relation, left-right relation and overlapping relation, root
Carry out effectively extracting Chinese character according to these three relations.
Further, it is as follows the step of structure Area Voronoi diagrams in the step (1):
Step (11), connected domain division is carried out to image first, each connected domain corresponds to the polygon of a closure, and right
Each connected domain and the background dot of connected domain is not divided into is marked;
Step (12), the profile for representing using Freeman coding strands each connected domain, and on profile and contoured interior
Pixel initialized, initial value is the mark of corresponding profile;
Step (13), the border for extending by the way of spread F reeman coding strands connected domain;Simultaneously on extended contour
Each point judged, if the point simultaneously also be located on another connected domain profile, then the pixel can be considered
One Voronoi boundary point, it is EDGE_POINT to set its value;Else if the point is a background dot, then its value is set
For the mark of corresponding profile;
Step (14), the extension of iteration, judgement, each connected domain of mark, when pixel all in plane all no longer wraps
When containing background pixel, otherwise represent that all pixels belong to some Voronoi area, otherwise it is the boundary point in region,
It this completes the structure of the Area Voronoi diagrams of image.
The principle of the present invention is:Firstly for an inscription on a tablet image of acquisition, image is built according to the method for extended boundary
Area Voronoi diagrams, each Voronoi in the figure contain an effective connected domain, the connected domain according to whether comprising
Chinese character can be divided into Chinese character connected domain and non-Chinese character connected domain, by certain criterion, merge Chinese character connected domain, go unless Chinese character connects
Logical domain, so as to orient the Chinese character region in an inscription on a tablet image.Extraction followed by individual Chinese character operates, equally the company of taking first
Each Chinese character in the method preliminary characterization region of logical domain analysis, but Chinese character included in the region now marked off and endless
It is whole, it is the result of adjacent Chinese characters combination.For further divided area, based on nine kinds of structures that Chinese character is common, conclude total
The relation of the boundary rectangle of three kinds of connected domains has been tied, has been respectively:Upper and lower relation, left-right relation and overlapping relation, utilize these three
Relation, the sub-dividing operations for carrying out next step with overlapping connected domain, so as to extract effective Chinese character.
Chinese character extraction research currently for an inscription on a tablet image is simultaneously few, deep in the present invention to analyze the Chinese in an inscription on a tablet image
The characteristic rule of word distribution, is the advantages of compared with the text extraction techniques of existing general pattern:
(1) the discrete sparse characteristic distributions of Chinese character in an inscription on a tablet image, are considered, by the side for building Area Voronoi diagrams
Formula divides to Chinese character region, while carries out region merging technique according to the skeleton width of Chinese character, can effectively, targetedly determine
Position goes out Chinese character region.
(2), in order to avoid the influence of noise, the smooth methods of TV are employed first to remove tiny in an inscription on a tablet image make an uproar
Sound point, then carries out binarization operation again.Simultaneously during individual character cutting, nine kinds of common Hanzi structure rules are carried out
Induction and conclusion, it is proposed that the segmentation algorithm based on connected domain and its bounding box, test result indicates that, the algorithm can be easy and effective
Extract individual Chinese character.
Brief description of the drawings
Fig. 1 shows the Area Voronoi diagrams constructed by an inscription on a tablet image in the present invention;
Fig. 2 shows the result figure after an inscription on a tablet image in Central Plains of the present invention and progress Chinese character zone location, wherein, figure (a) is
Former an inscription on a tablet image, figure (b) are to carry out the result figure after Chinese character zone location;
Fig. 3 (a), (b), (c), (d), (e), (f), (g), (h), (i) show to conclude nine kinds of common Chinese characters in the present invention
Structure;
Fig. 4 shows the definition to connected domain bounding box overlapping relation in the present invention;
Fig. 5 shows the false code of individual character segmentation algorithm in the present invention;
Fig. 6 shows the Chinese character segmentation result figure of an inscription on a tablet image in the present invention, wherein figure (a) is an inscription on a tablet image, figure (b) is upright stone tablet
The Chinese character segmentation result figure of texts and pictures picture;
Fig. 7 shows the Chinese character segmentation result figure of noisy an inscription on a tablet image in the present invention, wherein figure (a) is noisy upright stone tablet
Texts and pictures picture, figure (b) are the Chinese character segmentation result figure of noisy an inscription on a tablet image;
Fig. 8 shows a kind of schematic diagram of Chinese character extracting method towards an inscription on a tablet of the present invention.
Embodiment
The embodiment of the present invention is described in detail with reference to accompanying drawing.
The implementation process of the present invention is largely divided into four steps:The structure of Area Voronoi diagrams, Voronoi area are closed
And, the column split of an inscription on a tablet image, individual character cutting.
Step 1: the structure of Area Voronoi diagrams, is broadly divided into six basic steps:
(1) region division is carried out to image first, obtains the connected domain of n non-overlapping copies, be designated as G={ g1,g2,...,
gn, the polygon of the corresponding closure of each connected domain, its mark value is k (k=1,2 ..., n).Simultaneously by image not by
The value for the background pixel being divided into connected domain is set to BACKGROUND_VALUE.
(2) be located at connected domain inside in each pixel, set its pixel value be the connected domain mark value k (k=1,
2,…,n)。
(3) profile of each connected domain is represented using Freeman coding strands, is designated as ICk.And by the point on profile
Pixel value is set to k+M, and wherein M is a constant.
(4) each connected domain g is initializedkState variable Fk:=true, wherein k=1,2 ..., n.
(5) for each connected domain gkIts state variable is judged, if Fk==false, then export the connected domain;If
Fk==true, is followed the steps below.
(5.1) according to Freeman coding strands ICk, obtain the extended contour chain EC of the profilek。
(5.2) for profile chain ICkOn each pixel, if its value is not EDGE_VALUE, its value is set to k
(the original k+M of substitution).
(5.3) for extended contour chain ECkEach pixel, compare the size of its value, if located between M and M+k,
So the pixel is a point on other connected domain profiles, is designated as a Voronoi edge point, and set the point
Value be EDGE_VALUE.Otherwise, if value is BACKGROUND_VALUE, it is a background dot to illustrate the point, sets its value
For corresponding profile value k+M.
(5.4) if the institute on extended contour is not a little background dot, F is madek==false.
(5.5)ICk:=ECk。
(6) structure of Area Voronoi diagrams is completed.
Step 2: Voronoi area merges.
As shown in figure 1, each connected domain in image is represented with a Voronoi area, but each region is according to being
No to be divided into Chinese character connected domain and non-Chinese character connected domain comprising Chinese character component, Chinese character zone location is to merge Chinese character connected domain, is gone
Unless Chinese character connected domain.The standard deviation of element width in adjacent connected domain is used in the present invention as measurement Liang Ge Voronoi areas
Whether domain belongs to Chinese character region, and if it is similar, whether need to merge later.
It is mainly comprised the following steps:
(1) Euclidean distance transformation matrix is established, obtains the skeleton of each connected domain, is then calculated each on same skeleton
Point is to the minimum range on connected domain border, the width using 2 times of average of the average of these distances as component in connected domain
thickness(gk), it is specially:
Wherein, skFor on skeleton a bit, n is the number put on skeleton, CkRepresent the profile of connected domain, min (d (si,
Ck)) represent point s on skeletonkTo the beeline on profile border.
(2) Voronoi area positioned at picture centre is chosen as prime area, calculates the connection adjacent with the connected domain
Standard deviation between domain.Wherein, any two connected domain giAnd gjBetween the standard deviation of element width be defined as:
(3) size of standard of comparison difference, due to standard deviation occur larger value be concentrated mainly on non-Chinese character connected domain it
Between, therefore set suitable threshold value, you can Chinese character connected domain is determined whether by the size of standard deviation.The definition of the threshold value
For:
Step 3: an inscription on a tablet image column split
Orient behind the Chinese character region in an inscription on a tablet image, it is necessary to carry out binary conversion treatment to image, it is convenient with entering ranks point
The operation cut.But due to itself there is certain noise in image, follow-up point of influence of noise can be caused by directly carrying out binaryzation
Result is cut, therefore pretreatment operation is first carried out to image using TV smoothing algorithms in the present invention, filters out partial noise in image
Binarization operation is carried out again.Why TV smoothing algorithm is chosen, therefore this method can be good at the structural information of reservation Chinese character,
The situation of other smoothing algorithm blurred pictures will not occur.
Arranged in a column direction with certain regularity according to an inscription on a tablet calligraphy work, using the method for histogram projection,
The number of picture black pixel in each row is counted, due to certain background area (i.e. white picture between two row Chinese characters be present
Element) so that occur local minimum in histogram projection, the phenomenon of " trough " is presented, the foundation that thus can be cut as row.
Simultaneously in order to avoid some Chinese characters ledge for cutting interference, using connected domain and Contour extraction two
Method that person is combined is handled.Main process is as follows:
(1) when a certain row run into target pixel points in cutting image, the image zooming-out near the point is connected accordingly
Domain, extract connected domain in this abscissa identical boundary point X.
(2) Contour tracing operation is carried out, with most short apart from linking objective pixel and boundary point X, by the most short profile
It is designated as splitting point sequence.
Step 4: individual character cutting.
The composition of Chinese character is analyzed, is mainly made up of basic stroke (point, horizontal, vertical, slash, right-falling stroke etc.), each stroke is mutual
Connection, form the part of the radical, alternatively referred to as Chinese character of Chinese character.By analyzing the relation in Chinese character between part, can obtain
To nine kinds of situations as shown in Figure 3, it is summarised as up-down structure (such as prosperous, morning etc.), tiled configuration (such as you, say), includes
Structure (such as figure, tired etc.), semi-surrounding structure (such as sentence, huge etc.)., can will be same when dividing connected domain when picture quality is poor
The different component clusterings of one Chinese character can be by the different component clusterings of different Chinese character to same company into different connected domains, while also
In logical domain, the situation of " adhesion " is caused., can be by nine kinds of parts of Chinese character using the boundary rectangle of connected domain as the feature of connected domain
Structure is reduced to connected domain and three kinds of relations of external matrix:Upper and lower relation, left-right relation and overlapping relation.
As shown in figure 4, illustrate two connected domain giAnd gjThere is overlapping relation, with (Li,Ui) and (Ri,Di) be respectively
giBoundary rectangle correspond to the coordinate representation in the upper left corner and the lower right corner.The related definition of connected domain is as follows:
Wide Width (gi)=Ri-Li
High Height (gi)=Di-Ui
The ratio of width to height
Area Square (gi)=Width (gi)×Height(gi)
Adjacent connected domain merges later related definition:
Width Uwidth (gi,gj)=max (Ri,Rj)-min(Li,Lj)
Height Uheight (gi,gj)=max (Di,Dj)-min(Ui,Uj)
The ratio of width to height
Overlapping width Owidth (gi,gj)=Width (gi)+Width(gj)-Uwidth(gi,gj)
Overlapping height Oheight (gi,gj)=Height (gi)+Height(gj)-Uheight(gi,gj)
Overlapping area Osquare (gi,gj)=Owidth (gi,gj)×Oheight(gi,gj)
Unit defined above is indicated with the number of pixel.
After the overlapping relation progress quantization means of connected domain, it is possible to carry out the behaviour of the individual character cutting of next step
Make.
After obtaining single-row Chinese character region, and each row image zooming-out connected domain therein, it is designated as Ck(k=1,2 ...,
M), wherein M is the number of connected domain.The complete individual Chinese character of expression that this M connected domain has, has plenty of, is different Chinese character
Combination between different parts merges the different components of same Chinese character, it is necessary to its further subdivision.Its main cutting is calculated
The false code of method is as shown in Figure 5.
Using above-mentioned implementation process, the Chinese character carried out simultaneously to muting an inscription on a tablet image and noisy an inscription on a tablet image carries
The operation taken, its result is as shown in Figures 6 and 7, the results showed that the present invention has certain validity and noise immunity.
Claims (2)
1. a kind of Chinese character extracting method towards an inscription on a tablet, it is characterised in that comprise the following steps:
Step (1), using extended boundary method build an inscription on a tablet image Area Voronoi diagrams, each Voronoi area table
Show a connected domain, whether each connected domain is according to wherein comprising Chinese character being divided into Chinese character connected domain and non-Chinese character connected domain;By
In the discreteness of Chinese character, the separation between individual Chinese character stroke can be caused during Area Voronoi diagrams are built, so as to
Belong in different connected domains, therefore only include the component of part Chinese character, referred to as Chinese character;
Step (2), according to the otherness between connected domain element width standard deviation, go unless Chinese character connected domain, and merge all
Chinese character connected domain, so as to orient the whole Chinese character region in an inscription on a tablet image;
Step (3), Chinese character area image progress total variation (TV, Total Variation) smooth operation is removed necessarily
Noise spot, binaryzation is carried out afterwards so that image only includes two kinds of pixels of black and white;Calculate the row projection Nogata in bianry image
Figure, according to the intermittent rule of an inscription on a tablet calligraphy column distribution, is partitioned into single-row Chinese character region;
Connected domain in step (4), extraction Chinese character image row, the connected domain include single complete Chinese character, or are different Chinese character
Different piece combination, it is therefore desirable to connected domain is recombinated;Analyze structural characteristics of Chinese character, by connected domain boundary rectangle it
Between relation be summarized as three kinds of situations:Upper and lower relation, left-right relation and overlapping relation, carried out extracting the Chinese according to these three relations
Word.
2. as claimed in claim 1 towards the Chinese character extracting method of an inscription on a tablet, it is characterised in that:Built in the step (1)
The step of Area Voronoi diagrams, is as follows:
Step (11), connected domain division being carried out to image first, each connected domain corresponds to the polygon of a closure, and to each
Connected domain and the background dot of connected domain is not divided into is marked;
Step (12), the profile for representing using Freeman coding strands each connected domain, and on profile and contoured interior picture
Vegetarian refreshments is initialized, and initial value is the mark of corresponding profile;
Step (13), the border for extending by the way of spread F reeman coding strands connected domain;Simultaneously to every on extended contour
One point is judged, if the point also is located on another connected domain profile simultaneously, then the pixel can be considered one
Voronoi boundary points, it is EDGE_POINT to set its value;Else if the point is a background dot, then it is phase to set its value
The mark for the profile answered;
Step (14), the extension of iteration, judgement, each connected domain of mark, when pixel all in plane is all no longer comprising the back of the body
During scene element, that is, represent that all pixels belong to some Voronoi area, or the boundary point for region, this completes
The structure of the Area Voronoi diagrams of image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510441259.4A CN104992176B (en) | 2015-07-24 | 2015-07-24 | A kind of Chinese character extracting method towards an inscription on a tablet |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510441259.4A CN104992176B (en) | 2015-07-24 | 2015-07-24 | A kind of Chinese character extracting method towards an inscription on a tablet |
Publications (2)
Publication Number | Publication Date |
---|---|
CN104992176A CN104992176A (en) | 2015-10-21 |
CN104992176B true CN104992176B (en) | 2018-04-06 |
Family
ID=54303989
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510441259.4A Active CN104992176B (en) | 2015-07-24 | 2015-07-24 | A kind of Chinese character extracting method towards an inscription on a tablet |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN104992176B (en) |
Families Citing this family (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN106503688A (en) * | 2016-11-17 | 2017-03-15 | 西安理工大学 | Writing brush word minimum bounding box extracting method based on wavelet Smoothing |
CN107527055B (en) * | 2017-08-04 | 2018-12-11 | 佛山市国方商标服务有限公司 | Image divides card processing method, device and image search method, device and system |
CN108647680B (en) * | 2018-04-28 | 2021-11-12 | 北京盒子鱼教育科技有限公司 | Image positioning frame detection method and device |
CN116152277B (en) * | 2023-03-10 | 2023-09-22 | 麦岩智能科技(北京)有限公司 | Map segmentation method and device, electronic equipment and medium |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103186787A (en) * | 2011-12-31 | 2013-07-03 | 廖志武 | Low-quality Chinese character primary skeleton extraction algorithm based on point cloud model |
CN103400132A (en) * | 2013-07-02 | 2013-11-20 | Tcl集团股份有限公司 | Method and device for character segmentation |
CN104408455A (en) * | 2014-11-27 | 2015-03-11 | 上海理工大学 | Adherent character partition method |
-
2015
- 2015-07-24 CN CN201510441259.4A patent/CN104992176B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103186787A (en) * | 2011-12-31 | 2013-07-03 | 廖志武 | Low-quality Chinese character primary skeleton extraction algorithm based on point cloud model |
CN103400132A (en) * | 2013-07-02 | 2013-11-20 | Tcl集团股份有限公司 | Method and device for character segmentation |
CN104408455A (en) * | 2014-11-27 | 2015-03-11 | 上海理工大学 | Adherent character partition method |
Non-Patent Citations (3)
Title |
---|
古籍手写汉字图像分割算法研究;朱雷;《中国博士学位论文全文数据库 信息科技辑》;20111215(第12期);第I138-62页 * |
基于数字图像的碑文文字提取及三维展示;吕继增;《中国优秀硕士学位论文全文数据库 信息科技辑》;20131215(第S2期);第I138-1268页 * |
手写体汉字分割的研究与应用;杨侠;《中国优秀硕士学位论文全文数据库 信息科技辑》;20120815(第08期);第I138-1030页 * |
Also Published As
Publication number | Publication date |
---|---|
CN104992176A (en) | 2015-10-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107146280B (en) | Point cloud building reconstruction method based on segmentation | |
CN105931295B (en) | A kind of geologic map Extracting Thematic Information method | |
CN101996401B (en) | Target analysis method and apparatus based on intensity image and depth image | |
CN110738207A (en) | character detection method for fusing character area edge information in character image | |
CN104992176B (en) | A kind of Chinese character extracting method towards an inscription on a tablet | |
CN104299263B (en) | A kind of method that cloud scene is modeled based on single image | |
CN110197153B (en) | Automatic wall identification method in house type graph | |
CN106228528B (en) | A kind of multi-focus image fusing method based on decision diagram and rarefaction representation | |
CN105631455A (en) | Image main body extraction method and system | |
CN105488758A (en) | Image scaling method based on content awareness | |
CN109033972A (en) | A kind of object detection method, device, equipment and storage medium | |
CN103186904A (en) | Method and device for extracting picture contours | |
CN110544300B (en) | Method for automatically generating three-dimensional model based on two-dimensional hand-drawn image characteristics | |
CN103218605A (en) | Quick eye locating method based on integral projection and edge detection | |
CN104408445A (en) | Automatic real-time human body detecting method | |
CN111507334A (en) | Example segmentation method based on key points | |
CN104157002A (en) | Color image texture force tactile reproduction method based on color transform space | |
CN106326872A (en) | Method for extracting gesture contour in augmented reality system | |
CN110060260A (en) | A kind of image processing method and system | |
CN101430789B (en) | Image edge detection method based on Fast Slant Stack transformation | |
CN108960012A (en) | Feature point detecting method, device and electronic equipment | |
CN105069766B (en) | A kind of an inscription on a tablet restorative procedure based on the description of Chinese character image contour feature | |
CN105069742B (en) | A kind of soft Chinese character change method based on writing feature editor | |
CN106895794A (en) | A kind of method and device for obtaining laser beam scan path | |
CN106780535A (en) | A kind of gray level image processing method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |