CN113297897B - Writing brush character auxiliary exercise method based on LR sequence and center axis information - Google Patents
Writing brush character auxiliary exercise method based on LR sequence and center axis information Download PDFInfo
- Publication number
- CN113297897B CN113297897B CN202110256678.6A CN202110256678A CN113297897B CN 113297897 B CN113297897 B CN 113297897B CN 202110256678 A CN202110256678 A CN 202110256678A CN 113297897 B CN113297897 B CN 113297897B
- Authority
- CN
- China
- Prior art keywords
- segment
- segments
- group
- adjacent
- central axis
- 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
- 238000000034 method Methods 0.000 title claims abstract description 46
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 56
- 238000004364 calculation method Methods 0.000 claims abstract description 25
- 230000011218 segmentation Effects 0.000 claims description 10
- 238000010606 normalization Methods 0.000 claims description 6
- 238000007781 pre-processing Methods 0.000 claims description 5
- 230000004927 fusion Effects 0.000 claims description 4
- 238000012545 processing Methods 0.000 claims description 4
- 238000003672 processing method Methods 0.000 claims description 2
- 238000012549 training Methods 0.000 claims description 2
- 230000007547 defect Effects 0.000 abstract description 6
- 238000012986 modification Methods 0.000 abstract description 2
- 230000004048 modification Effects 0.000 abstract description 2
- 230000000007 visual effect Effects 0.000 abstract description 2
- 238000010586 diagram Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005305 interferometry Methods 0.000 description 2
- 238000012423 maintenance Methods 0.000 description 2
- 230000001737 promoting effect Effects 0.000 description 2
- 230000019771 cognition Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000002452 interceptive effect Effects 0.000 description 1
- 239000003550 marker Substances 0.000 description 1
- 239000000463 material Substances 0.000 description 1
Classifications
-
- 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
- G06V30/32—Digital ink
- G06V30/333—Preprocessing; Feature extraction
- G06V30/347—Sampling; Contour coding; Stroke extraction
-
- 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
-
- 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
- G06V30/24—Character recognition characterised by the processing or recognition method
- G06V30/242—Division of the character sequences into groups prior to recognition; Selection of dictionaries
- G06V30/244—Division of the character sequences into groups prior to recognition; Selection of dictionaries using graphical properties, e.g. alphabet type or font
-
- 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
- G06V30/32—Digital ink
- G06V30/36—Matching; Classification
- G06V30/387—Matching; Classification using human interaction, e.g. selection of the best displayed recognition candidate
-
- G—PHYSICS
- G09—EDUCATION; CRYPTOGRAPHY; DISPLAY; ADVERTISING; SEALS
- G09B—EDUCATIONAL OR DEMONSTRATION APPLIANCES; APPLIANCES FOR TEACHING, OR COMMUNICATING WITH, THE BLIND, DEAF OR MUTE; MODELS; PLANETARIA; GLOBES; MAPS; DIAGRAMS
- G09B11/00—Teaching hand-writing, shorthand, drawing, or painting
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Multimedia (AREA)
- Business, Economics & Management (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Image Analysis (AREA)
- Character Input (AREA)
Abstract
The invention discloses a writing brush character auxiliary exercise method based on an LR sequence and center axis information. The invention combines the simple polygon axle wire calculation method based on the LR sequence and the grassfire algorithm, utilizes the advantage of higher axle wire information accuracy of the polygon axle wire algorithm based on the LR sequence and the advantage of more visual skeleton wire main body part of the grassfire algorithm, overcomes the defect that the polygon axle wire algorithm based on the LR sequence is easily influenced by boundary noise and the defect that the skeleton wire information accuracy obtained by the grassfire algorithm is lower and has intermittent, and corrects the skeleton wire information obtained by the calculation of the former by using the accurate axle wire information obtained by the calculation of the former to obtain the more accurate Chinese brush character axle wire which is highly consistent with the center point track of the writing brush. The invention compares the information of the middle axes of the handwriting Chinese character and the template Chinese character, and proposes a modification suggestion, thereby being helpful for a beginner to learn the writing of the Chinese character.
Description
Technical Field
The invention relates to the engineering field of computer digital images and processing thereof, relates to real-time identification of a target object, in particular to a grassfire algorithm-based central axis calculation method and an LR sequence-based polygonal central axis calculation method.
Background
The method belongs to the category of 'protection and inheritance of Chinese traditional culture writing brush characters'. The writing brush calligraphy is used as a traditional artistic form and records the history of the development and evolution of Chinese characters from ancient times to the past. The writing of learning writing brush characters is an important way for promoting the excellent traditional culture of China and is an important way for improving personal intelligence and maintenance. The regular script handwriting of the writing brush is a necessary lesson maintenance for beginners. The training teacher is more focused on the "spirit" of the writing brush word, and the beginner is in the stage of learning the "shape" and cannot comprehend the "spirit". Note that the trend of the writing brush style (also called as middle style) is very similar to the central axis of the writing brush style, the method aims to quantify the position and the degree of the stroke deviation of the practicer by calculating the central axis of the writing brush style written by the user and comparing the central axis with the central axis of the writing brush style in the copybook, so that the beginner of the writing brush style is assisted to learn handwriting, and the aim of protecting and promoting inheritance and development of non-material cultural heritage by using modern technological means is fulfilled.
Disclosure of Invention
The invention aims at overcoming the defects of the prior art and provides a writing brush word auxiliary exercise method based on LR sequences and medial axis information. Because the trend of the strokes of the Chinese brush word (also called as the middle strokes) is very similar to the central axis of the Chinese brush word, the invention aims to assist the user in learning handwriting by calculating the central axis information of the Chinese brush word written by the user and comparing the central axis information of the Chinese brush word template in the copybook library, quantifying the deviation (comprising the position and the angle of the deviation) of the central axis information and providing a proposal.
The technical scheme provided by the invention comprises the following specific steps:
step 1: the user converts the written Chinese brush character into a digital image which can be stored and processed by a computer through digital equipment (such as mobile phone photographing), uploads the digital image to the computer, and searches a template Chinese brush character image of the Chinese brush character in a corresponding copybook library of the computer.
Step 2: preprocessing the handwriting brush character image and the template brush character image, including binarization, denoising, position normalization and size normalization.
Step 3: and extracting the boundary of the word in the Chinese brush word image and taking the boundary as vertex data. The first data in the vertex data is extracted as the first vertex, and all data in the vertex data are extracted for word boundaries on a time-by-time basis with the first vertex as a reference, so that the data in the vertex data are ordered.
Step 4: the vertex data is used to obtain the edge omega of a polygon P formed by the vertex and the straight line segment, wherein two end points of the straight line segment are respectively two adjacent vertices. And solving the accurate central axis information corresponding to the polygon P by using an algorithm based on an LR sequence.
Step 4.1: initializing to obtain a central axis segmentation sequence theta, wherein the central axis segmentation sequence theta comprises angular bisectors of each internal angle smaller than 180 degrees in the polygon P in an initialized state; each segment in the central axis segment sequence θ is ordered according to the clockwise order of the vertices to which it belongs on the boundary of the polygon P, and then the end point endPoint of each segment is calculated.
Step 4.2: and (3) setting the clockwise direction as the positive direction and the anticlockwise direction as the negative direction, determining LR marks of each segment by utilizing the intersection condition and endPoint of each segment and adjacent segments in the positive direction and the negative direction, and applying a self-growing algorithm to all segments with LR marks of 10 in the central axis segment sequence theta until no segment with LR marks of 10 is in the central axis segment sequence theta.
Step 4.3: all segments within the center axis segment sequence θ are grouped according to their LR markers, with the LR markers arranged to conform to the regular expression (r+) (l+) and the segments grouped into a set.
Step 4.4: after the grouping is finished, each group of segmented sequences is internally provided with a growing algorithm, and an interference processing algorithm is executed when interference occurs in the process of executing the growing algorithm;
the internal execution growth algorithm is specifically as follows: firstly, applying a joint growth algorithm to each unique RL pair, and marking the RL pair as a root pair; after two segments grow, a new segment is obtained, LR marks of the new segment are calculated according to the intersection condition of the new segment and adjacent segments in the positive direction and the negative direction and the endPoint of the new segment, RL pairs participating in growth are removed from the group, and the new segment is added into the group sequence; if the LR for the new segment is marked 10, then a self-growing algorithm is performed on it. And then, circularly performing an internal execution growth algorithm until the new segment obtained by growth does not have other segments in the positive direction or the negative direction in the group, or all the segments remained in the central axis sequence theta are intersected at one point.
Step 4.5: if a plurality of groups exist, executing step 4.4 on the segment sequence of each group until the new segment obtained by growth does not exist other segments in the positive direction or the negative direction of the new segment in the group, marking the new segment as B, if B does not exist adjacent segments in the positive direction group and exists adjacent segments in the negative direction group, judging the LR marks of adjacent segments outside the positive direction group of B again, if the LR marks of adjacent segments outside the group are changed into L, adding the adjacent segments outside the group into the group where B is located, and continuously executing step 4.4; if B has adjacent segments in the positive direction group and does not have adjacent segments in the negative direction group, judging the LR marks of the adjacent segments outside the negative direction group of B again, if the LR marks of the adjacent segments outside the group are changed into R, adding the adjacent segments outside the group into the group where B is located, and continuously executing the step 4.4; if B does not have adjacent segments in the positive direction group and does not have adjacent segments in the negative direction group, judging LR marks of adjacent segments outside the positive direction group and adjacent segments outside the negative direction group of B again, and if the LR marks of the adjacent segments outside the positive direction group are changed into L or the LR marks of the adjacent segments outside the negative direction group are changed into R, all the changed adjacent segments outside the group need to be added into the group where B is located, and continuously executing the step 4.4; if B does not have adjacent segments in the positive direction group or adjacent segments in the negative direction group, and no new out-of-group adjacent segments are added to the group in which B is located, B stops growing, and a growing algorithm is executed on other groups. If only one packet remains at this time, step 4.4 is performed on this packet until all the segments within this packet are intersected by the same point.
Step 4.6: if the growing algorithm is executed on all the groups and the growth of each group is stopped, when the rest of the segments are found not to cross the same point, the step returns to the step 4.3, and if the rest of the segments cross the same point, the calculation of the accurate central axis is finished.
Step 5: and (3) performing center axis simplification on the obtained accurate center axis to obtain the needed center axis information data.
Step 6: and solving skeleton line information corresponding to the Chinese brush character image by utilizing an improved grassfire algorithm thought.
Step 6.1: and inputting the preprocessed brush pen word image. And traversing all pixel points of the Chinese brush character image, marking the coordinates of the black pixel points as 1, and marking the rest pixel points as 0.
Step 6.2: let a=2, extract the edge of the word in the brush word image, change its mark to a.
Step 6.3: traversing all pixel points of the Chinese brush character image, finding out all pixel points marked as a, and modifying the mark of the pixel point marked as 1 in 8 adjacent areas of the pixel point to be a+1; the pixel marked as 0 is not processed; after the traversal of all the pixel points is completed, enabling a=a+1, and returning to the step 6.3 to carry out the next traversal if the marks of the pixel points are changed in the traversal process; if no pixel mark changes, the process proceeds to step 6.4.
Step 6.4: and traversing all the pixel points of the image, marking the mark of the pixel point which is traversed at present as i, and marking the pixel point as a skeleton point if the mark of 5 or more pixel points in 8 adjacent areas of the pixel point is i-1. All the marked skeleton points form skeleton lines of the Chinese brush word.
Step 7: and (3) fusing the axle wire information obtained in the step (5) with the skeleton wire information obtained in the step (6) to obtain the final required axle wire image of the Chinese brush character.
The accuracy of accurate central axis information calculated by a polygon central axis algorithm based on an LR sequence is higher, but the accurate central axis information is easily influenced by boundary noise to generate more segments which do not meet the demands of people; the skeleton line information obtained by the grassfire algorithm is low in accuracy and intermittent, but meets the requirements of people on the whole skeleton. Therefore, the images of the two are fused, the part of the skeleton line, which is missing and interrupted, in the skeleton line information is repaired by using the accurate central axis information, unnecessary miscellaneous points and miscellaneous edges are removed, and the final required central axis image of the Chinese brush character is obtained.
The fusion of the axle wire information and the skeleton wire information is as follows: the end points on the skeleton line (hereinafter skeleton points) are intermittent and discrete, while the precise central axis is continuous. Based on skeleton lines, a continuous precise central axis is divided into a plurality of sections by discrete skeleton points. After division, if two endpoints of a section of accurate central axis line segment are skeleton points, the section of accurate central axis line segment is reserved. As shown in fig. 12 (a), black line segments in the drawing are precise central axes, gray end points are skeleton points, continuous precise central axes are divided by the skeleton points, and only precise central axis segments with both end points being skeleton points are reserved, so that fig. 12 (b) is obtained.
Step 8: displaying central axis information I corresponding to the handwriting pen character image by using one color; and displaying the axle wire information II of the original Chinese brush character image by using another color, and adding the axle wire information I and the axle wire information II into the Chinese brush character image.
Step 9: and calculating the deviation between the central axis of the handwriting Chinese character and the central axis of the template Chinese character, and counting error results, wherein the position of the deviation (which strokes have deviation) and the size of the deviation (such as the angle of the deviation) are displayed by using the character information.
Some concepts and definitions
Definition of segments: assuming that the edge of the polygon P is Ω, Ω is composed of vertices and straight-line segments, a shortest path from any point inside Ω (inside the polygon P) to Ω necessarily intersects Ω at one point, denoted as point a, the inside of P is divided into a plurality of areas, each area is composed of all points that leave the polygon through the shortest path and pass through the same line segment or the same vertex on Ω, and the edges of these areas are composed of straight-line segments and parabolas, which are called segments.
Segmented trajectories: for points inside each region, exiting the polygon in the shortest path necessarily passes through the same straight line segment or the same vertex on Ω, which is collectively referred to as the feature edge of this region. For each segment, since each segment is the boundary between two regions, each region has a corresponding feature edge, which are referred to as the trajectory of the segment.
Determining L, R label: for each segment, determining whether an intersection point exists before the endPoint of the adjacent segment in the positive direction and the negative direction of the segment, and if so, judging which segment intersects the adjacent segment in the positive direction and the negative direction first.
If the segment is intersected with the adjacent segment in the positive direction only and is not intersected with the adjacent segment in the negative direction; or simultaneously intersects the positive and negative direction adjacent segments, but the straight line distance between the intersection point of the segment and the positive direction adjacent segment and the segment start point is smaller than the straight line distance between the intersection point of the segment and the negative direction adjacent segment and the segment start point, and the LR mark is determined as R. If the segment is intersected with the adjacent segment in the negative direction only and is not intersected with the adjacent segment in the positive direction; or simultaneously intersects the positive and negative direction adjacent segments, but the straight line distance between the intersection point of the segment and the positive direction adjacent segment and the segment start point is greater than the straight line distance between the intersection point of the segment and the negative direction adjacent segment and the segment start point, and the LR mark is determined to be L.
As shown in fig. 6 (a), the negative-direction neighboring segment of segment P is segment O, the positive-direction neighboring segment of segment P is segment Q, and segment P intersects only segment Q, so the LR flag of segment P is determined as R. As shown in fig. 6 (b), the negative-direction neighboring segment of segment P is segment O, the positive-direction neighboring segment of segment P is segment Q, the intersection point of segment P and segment O is N, and the intersection point of segment P and segment Q is M. Segment P intersects both segment O and segment Q, but the distance from the start of segment P to intersection point M is less than the distance from the start of segment P to intersection point N, so the LR signature of segment P is determined as R.
If the segment is not intersected with the adjacent segments in the positive direction and the negative direction, determining the LR mark of the segment as 10 or 3 according to the rule, wherein the segment with the LR mark of 10 needs to execute a self-growing algorithm, and the segment with the LR mark of 3 needs to wait for the other segments to finish growing first.
As shown in fig. 5, the rule that the segmentation markers are determined to be 10 or 3 is as follows: the segmented LR flag is determined to be 10 only when beforesitation and aftersitura are both 2. When one of BeforeStation or AfterSitation is 0 and the segment is not interfering, the LR flag for the segment is determined to be 3.
As shown in fig. 4, beforesitation refers to the relationship marks of a segment and its negative-direction neighboring segment, and aftersitation refers to the relationship marks of a segment and its positive-direction neighboring segment.
The sectional joint growth method comprises the following steps: and (3) sequentially carrying out joint growth on two segments marked as R and L according to a clockwise direction, wherein the two segments are intersected at the same point, a new segment is obtained by joint growth of the two segments, a start point of the new segment is an intersection point of the two segments, an end point of the new segment is obtained by independent calculation, a negative direction track of the new segment is a negative direction track of the segment marked as R, and a positive direction track of the new segment is a positive direction track of the segment marked as L.
The segmented self-growth method comprises the following steps: the segment marked as 10 is self-grown to obtain a new segment, the starting point of the new segment is the end point of the segment marked as 10, and the end point of the new segment is independently calculated. Only parabolic segments are obtained when they self-grow, and only linear segments are obtained when they self-grow. Of the two tracks of the new segment obtained by self-growth, one track of the new segment is identical to one track of the old segment, and the other track of the new segment is adjacent to the other track of the old segment on a polygon boundary.
As shown in fig. 7 (a), the linear segment P has a trajectory of a segment AB and a segment AC, and after P grows itself, a new parabolic segment P 'is obtained, and the trajectory corresponding to P' is a segment AB and a vertex C. As shown in fig. 7 (b), the trace of the parabolic segment P is a segment AB and a vertex C, and after P grows itself, a new straight segment P 'is obtained, and the trace corresponding to P' is a segment AB and a segment CD.
The interference processing method comprises the following steps: if each segment is considered as a node, the central axis resulting from the growth of many segments can be considered as a tree, and if a segment, whose positive and negative directions are considered as nodes, intersects with its offspring nodes of the segment nodes in either its positive or negative direction, interference occurs. Assuming segment A interferes with another segment B beside it and intersects with the B's descendant node segment C, it follows from step 8 that the growth starts with the center-most RL pair of each packet, i.e., the root pair, which is denoted as the core node of the tree. If segment C is not the core node, then changing the LR flag of this node of segment C and cancelling all ancestor node segments thereof, continuing the growing process; if the segment C is a core node, all ancestor nodes are revoked, and if the mark of the segment C is L, all marks of LR marked as R in the packet are changed into L, and the mark of the segment C is changed into R; if the label of C is R at this time, then all LR labels L in the group are changed to R, and the label of C is changed to L, and then the growth process is continued.
As shown in fig. 8, the LR markers for segment 1 through segment 5 are RLLRLs, respectively, where segment 1 and segment 2 are core nodes of the first packet. From the figure, it can be seen that segment 8 is grown from segment 4 and segment 5 in combination, segment 8 intersects with descendant node 3 of its adjacent segment 7 in the negative direction, and interference occurs. Since segment 3 is not the first grouped core node, then the LR flag for segment 3 is changed from L to R and all ancestor nodes thereof are revoked, i.e., segment 7 is revoked, and the growth process continues.
As shown in fig. 9, LR labels of segment 1 through segment 5 are RRLRLs, respectively, where segment 2 and segment 3 are core nodes of the first packet. From the figure, it can be seen that segment 8 is grown from segment 4 and segment 5 in combination, segment 8 intersects with descendant node 3 of its adjacent segment 7 in the negative direction, and interference occurs. Since segment 3 is the core node of the first packet, then all ancestor nodes of segment 3, namely segment 6 and segment 7, are revoked, since the LR flags of segment 3 are L, then the LR flags of segment 1 and segment 2 are changed from R to L, and the LR flag of segment 3 is changed to R, and then the growth process continues.
The invention has the beneficial effects that:
the invention has the core technical characteristics of high real-time performance of center axis calculation and high rationality of center axis information. According to the simple polygon central axis calculation method based on the LR sequence, which is provided by the technical scheme of the invention, the structure of a central axis self tree and the intersection condition of each segment and adjacent segments in the positive direction and the negative direction are utilized, the characteristic that two segments grow into one segment each time is utilized, one segment is reduced in each joint growth, and experiments show that the algorithm can calculate the accurate central axis of the polygon in linear time and can obtain better calculation instantaneity.
The invention combines the simple polygon axle wire calculation method based on the LR sequence and the grassfire algorithm, utilizes the advantage of higher axle wire information accuracy of the polygon axle wire algorithm based on the LR sequence and the advantage of more visual skeleton wire main body part of the grassfire algorithm, overcomes the defect that the polygon axle wire algorithm based on the LR sequence is easily influenced by boundary noise and the defect that the skeleton wire information accuracy obtained by the grassfire algorithm is lower and has intermittent, and corrects the skeleton wire information obtained by the calculation of the former by using the accurate axle wire information obtained by the calculation of the former to obtain the more accurate Chinese brush character axle wire which is highly consistent with the center point track of the writing brush. Meanwhile, the invention compares the information of the middle axes of the handwriting Chinese character and the template Chinese character, and proposes a modification suggestion, so that a user can intuitively find out the defects of the user, and the invention has a preliminary cognition on how to improve, thereby being helpful for a beginner to learn the writing of the Chinese character.
The method is easy to implement, ingenious in conception and high in calculation efficiency for the central axis of the Chinese brush word, is a product of combining the modern information technology with the traditional culture, provides a new digital auxiliary exercise mode for the traditional Chinese calligraphy culture, and is beneficial to inheritance and development of the Chinese brush word culture in the modern society.
Drawings
FIG. 1 is a flow chart of the overall algorithm;
FIG. 2 is a flow chart of calculating accurate center axis information based on a polygon center axis calculation method of LR sequences;
FIG. 3 is a flow chart of the Grassfire algorithm for calculating skeleton line information of a Chinese brush word image;
FIG. 4 is a table of determining the relationship between a segment P and its neighboring segments in a polygon center axis calculation method based on an LR sequence;
FIG. 5 is a segmented LR marker determination table for a polygon centerline axis calculation method based on an LR sequence;
FIG. 6 is an example graph of determining whether an LR flag is L or R when a segment intersects an adjacent segment;
FIG. 7 is a schematic illustration of a self-grown new segmented trajectory;
FIG. 8 is a schematic diagram of an interferometry (non-root pair);
FIG. 9 is a schematic diagram of an interferometry (root pair);
FIG. 10 is an example diagram of grouping a sequence of center axis segments;
FIG. 11 is a schematic diagram of a grassfire algorithm;
FIG. 12 is a schematic diagram showing the fusion of the central axis information and the skeleton line information;
FIG. 13 is a simplified central axis information of the "C" character passing through the central axis;
fig. 14 shows the skeleton line information obtained by the grassfire algorithm in the "c" word;
fig. 15 is a final central axis image of the "c" word.
Detailed Description
The invention will be described in further detail with reference to the accompanying drawings and examples of embodiments:
As shown in fig. 1 and fig. 2, the implementation process of the writing brush word auxiliary exercise method based on LR sequence and medial axis information of the present invention is mainly divided into five processes:
firstly, inputting and preprocessing a handwriting Chinese character image and a template Chinese character image;
obtaining accurate central axis information of the Chinese brush character image by using a polygon central axis calculation method based on an LR sequence;
thirdly, calculating to obtain skeleton line information of the brush pen word image by using a grassfire algorithm;
step four, combining accurate axle wire information and skeleton line information of the Chinese brush character image to obtain a final axle wire image;
and fifthly, comparing the axial line images of the handwriting Chinese character with the template Chinese character, and outputting a comparison result and a suggestion.
The first process, the input and pretreatment of the handwriting Chinese character image and the template Chinese character image mainly comprises two basic steps:
(1) The user converts the written Chinese brush character into a digital image which can be stored and processed by a computer through digital equipment (such as mobile phone photographing), and uploads the digital image to the computer, and searches a template Chinese brush character image corresponding to the Chinese brush character in a corresponding copybook library of the computer through a manual font input mode.
(2) Preprocessing the handwriting brush character image and the template brush character image. The original Chinese brush character image is converted into a gray level image, and an appropriate threshold value is selected according to the specific condition of the image to binarize the image. And denoising the brush pen word image to remove the foreign points. Traversing each pixel value represented by the image pixel, and changing the mark of the pixel into 0 if the mark of the pixel is 1 and the marks of 6 or more pixels in 8 adjacent areas of the pixel are 0; if the pixel has a mark of 0 and 6 or more pixels in its 8 neighbors have a mark of 1, the pixel is marked as 1. And performing position normalization on the binary and denoising Chinese brush character image, namely cutting off redundant parts of the Chinese brush character image by taking the uppermost, lowermost, leftmost and rightmost pixel points of the stroke pixel points as rectangular boundaries. And then carrying out size normalization on the two brush word images processed by the method, namely adjusting the two brush word images to the same size.
The second process, which is shown in fig. 2 and mainly comprises nine steps, is to calculate and obtain accurate central axis information of the Chinese brush character by using a polygon central axis calculation method based on an LR sequence:
(1) The black pixel is noted as 1 and the white pixel is noted as 0. Traversing all pixel points of the preprocessed Chinese brush character image, and if the mark of the pixel point is 1 and the pixel point with the mark of 0 exists in the 8 neighborhood of the pixel point, taking the coordinates of the pixel point with the mark of 1 as boundary vertex data record.
(2) And extracting first data in the vertex data as a first vertex, and sequentially arranging the vertices in a clockwise order by taking the first vertex as a reference to obtain a boundary omega of a polygon P formed by the vertices and a straight line segment, wherein two endpoints of the straight line segment are two adjacent vertices respectively.
(3) Initializing to obtain a central axis segmentation sequence theta, wherein the central axis segmentation sequence theta comprises angular bisectors of each inner angle smaller than 180 degrees in the polygon P in an initialized state, each segment in the central axis segmentation sequence theta is ordered according to the clockwise sequence of the vertex of the segment on the boundary omega of the polygon P, and then the end point endPoint of each segment is calculated.
(4) And (3) setting the clockwise direction as the positive direction and the anticlockwise direction as the negative direction, determining the LR mark of each segment by utilizing the intersection condition and the endPoint of each adjacent segment in the positive direction and the negative direction of each segment, and applying a self-growing algorithm to the segment with the LR mark of 10 in the central axis segment sequence theta until no segment with the LR mark of 10 in the central axis segment sequence theta.
(5) All segments within the central axis segment sequence θ are grouped according to their LR markers, which align with the segment sequence of regular expression (r+) (l+) and are grouped into a set.
As shown in fig. 10, all segments in the central axis segment sequence θ have LR markers RLLLLRLRRRLRRRLLLL in order, and they need to be divided into RLLLL, RL, RRRL, RRRLLLL four groups according to the regular expression (r+) (l+).
(6) Executing a growing algorithm in the segmented sequence of each group which is currently grouped, and executing an interference processing algorithm when interference occurs in the process of executing the growing algorithm;
the internal execution growth algorithm is specifically as follows, firstly, a joint growth algorithm is applied to each unique RL pair, and the RL pair is recorded as a root pair; after two segments grow, a new segment is obtained, LR marks of the new segment are calculated according to the intersection condition of the new segment and adjacent segments in the positive direction and the negative direction and the endPoint of the new segment, RL pairs participating in growth are removed from the group, and the new segment is added into the group sequence; if the LR for the new segment is marked 10, then a self-growing algorithm is performed on it. And then the growth algorithm is circularly executed in the group until the new segment obtained by growth does not have other segments in the positive direction or the negative direction in the group, or all the segments remained in the central axis sequence theta are intersected at one point.
(7) If a plurality of groups exist, executing a step (6) on the segment sequence of each group until the new segment obtained by growth does not exist other segments in the positive direction or the negative direction of the new segment in the group, marking the new segment as B, if B does not exist adjacent segments in the positive direction group and exists adjacent segments in the negative direction group, judging LR marks of adjacent segments outside the positive direction group of B again, and if the LR marks of adjacent segments outside the group are changed to L, adding the adjacent segments outside the group into the group where B is located, and continuing to execute the step (6); if the adjacent segments in the positive direction group exist and the adjacent segments in the negative direction group do not exist, the LR marks of the adjacent segments outside the negative direction group of the B are re-judged, if the LR marks of the adjacent segments outside the group are changed into R, the adjacent segments outside the group are added into the group where the B is located, and the step (6) is continuously executed; if B does not have adjacent segments in the positive direction group and does not have adjacent segments in the negative direction group, judging LR marks of adjacent segments outside the positive direction group and adjacent segments outside the negative direction group of B again, and if the LR marks of the adjacent segments outside the positive direction group are changed into L or the LR marks of the adjacent segments outside the negative direction group are changed into R, all the changed adjacent segments outside the group need to be added into the group where B is located, and continuing to execute the step (6); if B does not have adjacent segments in the positive direction group or adjacent segments in the negative direction group, and no new out-of-group adjacent segments are added to the group in which B is located, B stops growing, and a growing algorithm is executed on other groups. If only one packet remains at this time, step (6) is performed on this packet until all the segments within this packet intersect at the same point.
(8) If all the groups are executed with the growth algorithm and each group stops growing, returning to the step (5) when the rest of the segments are found not to cross the same point, ending the calculation if the rest of the segments cross the same point, and obtaining the tree structure consisting of the initial segments and all the segments obtained by growing, namely the accurate central axis of the polygon P.
(9) And (3) carrying out center axis simplification on the accurate center axis of the obtained polygon P to obtain the required data of the accurate center axis information.
The center shaft is simplified: in the growth process of the central axis segment, all angular bisector segments in the central axis segment sequence theta obtained by initialization are marked as leaf segments. The new segment obtained by the self-growth of the leaf segment is marked as a leaf segment, the new segment obtained by the self-growth of the non-leaf segment is marked as a non-leaf segment, and the new segment obtained by the joint growth is marked as a non-leaf segment. When the joint growth of two leaf segments occurs, the two leaf segments are marked as segments A and B, the total length of the segment A and its descendant segments and the total length of the segment B and its descendant segments are calculated, and the leaf segment with the longer total length and its descendant segments are marked as non-leaf segments.
When interference occurs, the segment added again to the central axis segment sequence theta after the interference treatment is marked as C, and if the mark of the segment C is a non-leaf segment but is originally a leaf segment, the segment C and the descendant segments thereof are re-marked as the leaf segments.
As shown in fig. 4, at the end of the accurate central axis calculation, 2-3 segments remain in the central axis segment sequence θ:
1) The remaining 2 segments: the 2 segments must then intersect or overlap at a point, where there is one leaf segment and one non-leaf segment. The leaf segment and its offspring segments are marked as non-leaf segments.
2) The remaining 3 segments: at this point the 3 segments must intersect at a point where there are two leaf segments and one non-leaf segment. The two leaf segments are marked as A and B, the total length of the leaf segment A and its offspring segment and the total length of the leaf segment B and its offspring segment are calculated, and the leaf segment with the longer total length and its offspring segment are marked as non-leaf segments.
The precise central axis segment marked as a leaf segment is hidden when the precise central axis information is output. Fig. 13 shows accurate central axis information of the "c" word after the central axis is simplified.
Thirdly, solving skeleton line information corresponding to the Chinese brush character image by using an improved grassfire algorithm idea, wherein the process is mainly divided into four steps as shown in fig. 3:
(1) And inputting the preprocessed brush pen word image. After the image is binarized, each pixel is represented by 8 bits, 0 represents black, and 255 represents white. And traversing all pixel points of the Chinese brush character image, marking the coordinate mark where the black pixel point is positioned as 1, and marking the rest pixel points as 0.
(2) Let a=2, extract the edge of the word in the brush word image, change its mark to a.
(3) Traversing all pixel points of the Chinese brush character image, finding out all pixel points marked as a, and modifying the mark of the pixel point marked as 1 in 8 adjacent areas of the pixel point to be a+1; the pixel marked 0 is not processed. After the traversal of all the pixel points is completed, enabling a=a+1, and returning to the beginning of the step (3) to carry out the next traversal if the marks of the pixel points are changed in the traversal process; if no mark of the pixel point changes, the process proceeds to (4).
As shown in fig. 11, the edges of the image are extracted, labeled a, where a=2. Traversing all pixel points marked with 2, and modifying all pixel points marked with 1 in the 8 neighborhood of the pixel point to a+1, namely modifying to 3; the pixel marked 0 is not processed. After one pass is completed, let a=a+1, where a=3. And the next round of traversal is performed because the marks of the pixel points are changed in the traversal process. Traversing all pixel points marked with 3, and modifying all pixel points marked with 1 in the 8 neighborhood of the pixel point to a+1, namely modifying to 4; the pixel marked 0 is not processed. After one pass is completed, let a=a+1, where a=4. And so on until no mark of the pixel point changes, and then the next step is performed.
(4) And traversing all the pixel points of the image, in the traversing process, setting the mark of the pixel point traversed at present as i, and if the mark of 5 or more pixel points in 8 adjacent areas of the pixel point is i-1, marking the pixel point as a skeleton point. All the marked skeleton points form skeleton lines of the Chinese brush word.
As shown in fig. 11, the pixels marked with yellow in the figure are skeleton points, and all the skeleton points constitute skeleton lines. Fig. 14 shows the skeleton line information in the "c" shape.
And step four, fusing the accurate central axis information obtained in the step two and the skeleton line information obtained in the step three to obtain a final required central axis image of the Chinese brush character, wherein the step mainly comprises two steps:
(1) And obtaining an accurate central axis image of the Chinese brush character by using a drawing program, and marking pixels where the accurate central axis is positioned.
(2) Repairing the part of the skeleton line which is missing and interrupted in the skeleton line information by using the accurate central axis information, and removing unnecessary miscellaneous points and miscellaneous edges to obtain the final required central axis image of the Chinese brush character.
The fusion of the axle wire information and the skeleton wire information is as follows: the end points on the skeleton line (hereinafter skeleton points) are intermittent and discrete, while the precise central axis is continuous. Based on skeleton lines, a continuous precise central axis is divided into a plurality of sections by discrete skeleton points. After division, if two endpoints of a section of accurate central axis line segment are skeleton points, the section of accurate central axis line segment is reserved.
Fig. 15 is a final central axis image of the "c" word.
Outputting a central axis image of the handwriting Chinese character and the template Chinese character, calculating central axis deviation and outputting a comparison result, wherein the process mainly comprises the following two steps:
(1) Displaying central axis information I corresponding to the handwriting pen character image by using one color (such as white); and displaying the axle wire information II of the template Chinese brush character image by using another color (such as red), adding the axle wire information I and the axle wire information II into the Chinese brush character image, and outputting the image.
(2) And calculating the deviation between the central axis of the handwriting brush character and the central axis of the template brush character, counting error results, and displaying the position of the deviation (which strokes have deviation) and the size of the deviation (such as the angle of the deviation) by using the character information.
Claims (8)
1. A writing brush word auxiliary exercise method based on LR sequence and center axis information is characterized by comprising the following steps:
step 1: the user converts the written brush calligraphy into a digital image which can be stored and processed by a computer through digital equipment, uploads the digital image to the computer, and searches a template brush calligraphy image of the brush calligraphy in a corresponding copybook library of the computer;
step 2: preprocessing the handwriting brush character image and the template brush character image, wherein the preprocessing comprises binarization, denoising, position normalization and size normalization;
Step 3: extracting the boundary of the character in the Chinese brush character image and taking the boundary as vertex data; the first data in the vertex data is extracted as the first vertex, all data in the vertex data are extracted for word boundaries in time by taking the first vertex as a reference, so that the data in the vertex data are ordered;
step 4: obtaining an edge omega of a polygon P formed by vertexes and straight line segments by using vertex data, wherein two end points of the straight line segments are two adjacent vertexes respectively; solving accurate central axis information corresponding to the polygon P by utilizing an algorithm based on an LR sequence;
step 5: the obtained accurate central axis is subjected to central axis simplification, and required central axis information data are obtained;
step 6: solving skeleton line information corresponding to the Chinese brush character image by utilizing an improved grassfire algorithm thought;
step 7: fusing the axle wire information obtained in the step 5 with the skeleton wire information obtained in the step 6 to obtain a final required axle wire image of the Chinese brush character;
step 8: displaying central axis information I corresponding to the handwriting pen character image by using one color; displaying the axle wire information II of the original Chinese brush character image by using another color, and adding the axle wire information I and the axle wire information II into the Chinese brush character image;
Step 9: calculating the deviation between the central axis of the handwriting Chinese character and the central axis of the template Chinese character, counting error results, and displaying the position of the deviation and the size of the deviation by using character information;
the step 4 is specifically realized as follows:
step 4.1: initializing to obtain a central axis segmentation sequence theta, wherein the central axis segmentation sequence theta comprises angular bisectors of each internal angle smaller than 180 degrees in the polygon P in an initialized state; each segment in the central axis segment sequence theta is ordered according to the clockwise sequence of the vertex of the central axis segment sequence theta on the boundary of the polygon P, and then the end point endPoint of each segment is calculated;
step 4.2: setting the clockwise direction as positive direction and the anticlockwise direction as negative direction, determining LR marks of each segment by utilizing the intersection condition and endPoint of each adjacent segment in the positive direction and the negative direction of each segment, and applying a self-growing algorithm to all segments with LR marks of 10 in the central axis segment sequence theta until no segment with LR marks of 10 in the central axis segment sequence theta;
step 4.3: grouping all the segments in the central axis segment sequence theta according to LR marks, wherein the LR marks are arranged to be in accordance with the regular expression (R+) (L+) and the segments are divided into a group;
step 4.4: after the grouping is finished, each group of segmented sequences is internally provided with a growing algorithm, and an interference processing algorithm is executed when interference occurs in the process of executing the growing algorithm;
Step 4.5: if a plurality of groups exist, executing step 4.4 on the segment sequence of each group until the new segment obtained by growth does not exist other segments in the positive direction or the negative direction of the new segment in the group, marking the new segment as B, if B does not exist adjacent segments in the positive direction group and exists adjacent segments in the negative direction group, judging the LR marks of adjacent segments outside the positive direction group of B again, if the LR marks of adjacent segments outside the group are changed into L, adding the adjacent segments outside the group into the group where B is located, and continuously executing step 4.4; if B has adjacent segments in the positive direction group and does not have adjacent segments in the negative direction group, judging the LR marks of the adjacent segments outside the negative direction group of B again, if the LR marks of the adjacent segments outside the group are changed into R, adding the adjacent segments outside the group into the group where B is located, and continuously executing the step 4.4; if B does not have adjacent segments in the positive direction group and does not have adjacent segments in the negative direction group, judging LR marks of adjacent segments outside the positive direction group and adjacent segments outside the negative direction group of B again, and if the LR marks of the adjacent segments outside the positive direction group are changed into L or the LR marks of the adjacent segments outside the negative direction group are changed into R, all the changed adjacent segments outside the group need to be added into the group where B is located, and continuously executing the step 4.4; if B does not have adjacent segments in the positive direction group or does not have adjacent segments in the negative direction group and no new out-of-group adjacent segments are added into the group where B is located, B stops growing, and a growing algorithm is executed on other groups; if only one packet is left at this time, step 4.4 is continuously performed on this packet until all the segments within this packet are intersected at the same point;
Step 4.6: if all the groups are executed with a growth algorithm and each group stops growing, returning to the step 4.3 when the rest of the segments are found not to cross the same point, and ending the calculation of the accurate central axis if the rest of the segments cross the same point;
the step 6 is specifically realized as follows:
step 6.1: inputting a preprocessed brush pen character image; traversing all pixel points of the Chinese brush character image, marking the coordinates of the black pixel points as 1, and marking the rest pixel points as 0;
step 6.2: setting a=2, extracting the edges of the characters in the Chinese brush character image, and changing the marks of the edges into a;
step 6.3: traversing all pixel points of the Chinese brush character image, finding out all pixel points marked as a, and modifying the mark of the pixel point marked as 1 in 8 adjacent areas of the pixel point to be a+1; the pixel marked as 0 is not processed; after the traversal of all the pixel points is completed, enabling a=a+1, and returning to the step 6.3 to carry out the next traversal if the marks of the pixel points are changed in the traversal process; if no mark of the pixel point changes, the step 6.4 is entered;
step 6.4: traversing all pixel points of the image, marking the pixel points which are traversed at present as i, and marking the pixel points as skeleton points if the marks of 5 or more pixel points in 8 adjacent areas of the pixel points are i-1; all the marked skeleton points form skeleton lines of the Chinese brush word.
2. The method for assisting in practicing Chinese brush characters based on LR sequences and medial axis information according to claim 1, wherein the internal execution growth algorithm in step 4.4 is specifically as follows: firstly, applying a joint growth algorithm to each unique RL pair, and marking the RL pair as a root pair; after two segments grow, a new segment is obtained, LR marks of the new segment are calculated according to the intersection condition of the new segment and adjacent segments in the positive direction and the negative direction and the endPoint of the new segment, RL pairs participating in growth are removed from the current group, and the new segment is added into the group sequence of the current group; if the LR flag of the new segment is 10, executing a self-growing algorithm on the LR flag of the new segment; and then, circularly performing an internal execution growth algorithm until the new segment obtained by growth does not have other segments in the positive direction or the negative direction in the group, or all the segments remained in the central axis sequence theta are intersected at one point.
3. The writing brush word auxiliary exercise method based on the LR sequences and the medial axis information according to claim 1 or 2, wherein the fusion of the medial axis information and the skeleton line information in the step 7 is as follows: because the skeleton points on the skeleton line are intermittent and discrete, the precise central axis is continuous; dividing a continuous accurate central axis into a plurality of sections by using a discrete framework point with a framework line as a reference; after division, if two endpoints of a section of accurate central axis line segment are skeleton points, the section of accurate central axis line segment is reserved; dividing the continuous accurate central axis by using skeleton points, and only keeping the accurate central axis sections with two end points being skeleton points;
The correlation of the segments is defined as follows:
definition of segments: assuming that the edge of the polygon P is omega, wherein omega consists of vertexes and straight line segments, a shortest path from any point in the omega to the omega is intersected with the omega at one point, the shortest path is marked as point A, the interior of the polygon P is divided into a plurality of areas, each area consists of all points which are separated from the polygon by the shortest path and pass through the same line segment or the same vertex on the omega, and the edges of the areas consist of straight line segments and parabolas, wherein the straight line segments or the parabolas are called segmentation;
segmented trajectories: for points inside each region, the polygon must pass through the same straight line segment or the same vertex on Ω by the shortest path, and the line segment or the vertex is collectively called the characteristic edge of the region; for each segment, since each segment is the boundary between two regions, each region has a corresponding feature edge, which are referred to as the trajectory of the segment.
4. A method of auxiliary training for brush calligraphy based on LR sequence and medial axis information according to claim 3, characterized in that the determination of L, R label is as follows: for each segment, determining whether an intersection point before the endPoint exists in the adjacent segments in the positive direction and the negative direction of the segment, if so, judging which segment intersects with the adjacent segments in the positive direction and the negative direction first;
If the segment is intersected with the adjacent segment in the positive direction only and is not intersected with the adjacent segment in the negative direction; or simultaneously intersecting the adjacent positive and negative segments, wherein the linear distance between the intersection point of the segment and the adjacent positive segment and the segment start point is smaller than the linear distance between the intersection point of the segment and the adjacent negative segment and the segment start point, and the LR mark is determined as R; if the segment is intersected with the adjacent segment in the negative direction only and is not intersected with the adjacent segment in the positive direction; or simultaneously intersecting the adjacent positive and negative segments, wherein the linear distance between the intersection point of the segment and the adjacent positive segment and the segment start point is greater than the linear distance between the intersection point of the segment and the adjacent negative segment and the segment start point, and the LR mark is determined to be L;
if the segment is not intersected with the adjacent segments in the positive direction and the negative direction, determining that the LR mark of the segment is 10 or 3 according to the rule, and executing a self-growing algorithm on the segment with the LR mark of 10, wherein the segment with the LR mark of 3 is required to wait for the other segments to finish growing first;
the rule that the segmentation markers are determined to be 10 or 3 is as follows: the segmented LR flag is determined to be 10 only when beforesitation and aftersitura are both 2; when one of Beforesitation or Aftersination is 0 and the segment is not interfered, the LR flag of the segment is determined to be 3;
BeforeStudication refers to the relationship labels of segments and their negative neighbors, and AfterSitation refers to the relationship labels of segments and their positive neighbors.
5. The writing brush word assisting exercise method based on the LR sequences and the medial axis information according to claim 4, wherein the writing brush word assisting exercise method is characterized in that:
the sectional joint growth method comprises the following steps: sequentially carrying out joint growth on two segments marked as R and L according to a clockwise direction, wherein the two segments are intersected at the same point, a new segment is obtained by joint growth of the two segments, a start point of the new segment is an intersection point of the two segments, an end point of the new segment is obtained by independent calculation, a negative direction track of the new segment is a negative direction track of the segment marked as R, and a positive direction track of the new segment is a positive direction track of the segment marked as L;
the segmented self-growth method comprises the following steps: self-growing the segment marked as 10 to obtain a new segment, wherein the starting point of the new segment is the end point of the segment marked as 10, and the end point of the new segment is obtained by independent calculation; when the linear segment grows by self, only the parabolic segment can be obtained, and when the parabolic segment grows by self, only the linear segment can be obtained; of the two tracks of the new segment obtained by self-growth, one track of the new segment is identical to one track of the old segment, and the other track of the new segment is adjacent to the other track of the old segment on a polygon boundary.
6. The method for assisting in practicing Chinese brush characters based on LR sequences and medial axis information according to claim 4, wherein the interference processing method is characterized by comprising the following steps: if each segment is regarded as a node, the central axis obtained by growing a plurality of segments is regarded as a tree, if a segment is regarded as a node, and the segments in the positive direction and the negative direction intersect with the descendant nodes of the segment nodes in the positive direction or the negative direction, then interference occurs; assuming segment A interferes with another segment B beside it and intersects with B's descendant node segment C, growing starting with the center-most RL pair of each packet, and marking the RL pair as the core node of the tree; if segment C is not the core node, then changing the LR flag of this node of segment C and cancelling all ancestor node segments thereof, continuing the growing process; if the segment C is a core node, all ancestor nodes are revoked, and if the mark of the segment C is L, all marks of LR marked as R in the packet are changed into L, and the mark of the segment C is changed into R; if the label of C is R at this time, then all LR labels L in the group are changed to R, and the label of C is changed to L, and then the growth process is continued.
7. The method for assisting in practicing Chinese brush characters based on LR sequences and medial axis information according to claim 4, wherein the medial axis is simplified: in the growth process of the axle wire section, firstly, marking all angular bisector sections in the axle wire section sequence theta obtained by initialization as leaf sections; marking a new segment obtained by self-growth of a leaf segment as a leaf segment, marking a new segment obtained by self-growth of a non-leaf segment as a non-leaf segment, and marking a new segment obtained by joint growth as a non-leaf segment; when the joint growth of two leaf segments occurs, the two leaf segments are marked as segments A and B, the total length of the segment A and its offspring segment and the total length of the segment B and its offspring segment are calculated, and the leaf segment with the longer total length and its offspring segment in the two segments are marked as non-leaf segments;
when interference occurs, the segment added again to the central axis segment sequence theta after the interference treatment is marked as C, and if the mark of the segment C is a non-leaf segment but is originally a leaf segment, the segment C and the descendant segments thereof are re-marked as the leaf segments.
8. The method for assisting in practicing Chinese brush characters based on LR sequences and medial axis information according to claim 4, wherein 2-3 segments remain in the medial axis segment sequence θ when the calculation of the precise medial axis is finished:
When 2 segments remain: the 2 segments must then intersect at a point or overlap, with one leaf segment and one non-leaf segment; marking the leaf segment and its offspring segment as non-leaf segment;
when the remaining 3 segments: at this point 3 segments must intersect at a point, with two leaf segments and one non-leaf segment; the two leaf segments are marked as A and B, the total length of the leaf segment A and its offspring segment and the total length of the leaf segment B and its offspring segment are calculated, and the leaf segment with the longer total length and its offspring segment are marked as non-leaf segments.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110256678.6A CN113297897B (en) | 2021-03-09 | 2021-03-09 | Writing brush character auxiliary exercise method based on LR sequence and center axis information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110256678.6A CN113297897B (en) | 2021-03-09 | 2021-03-09 | Writing brush character auxiliary exercise method based on LR sequence and center axis information |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113297897A CN113297897A (en) | 2021-08-24 |
| CN113297897B true CN113297897B (en) | 2024-03-01 |
Family
ID=77319194
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110256678.6A Active CN113297897B (en) | 2021-03-09 | 2021-03-09 | Writing brush character auxiliary exercise method based on LR sequence and center axis information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113297897B (en) |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104715256A (en) * | 2015-03-04 | 2015-06-17 | 南昌大学 | Auxiliary calligraphy exercising system and evaluation method based on image method |
| CN106980857A (en) * | 2017-02-24 | 2017-07-25 | 浙江工业大学 | A kind of Brush calligraphy segmentation recognition method based on rubbings |
| CN107742309A (en) * | 2017-10-20 | 2018-02-27 | 杭州电子科技大学 | The axis computational methods of simple polygon based on LR sequences |
-
2021
- 2021-03-09 CN CN202110256678.6A patent/CN113297897B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104715256A (en) * | 2015-03-04 | 2015-06-17 | 南昌大学 | Auxiliary calligraphy exercising system and evaluation method based on image method |
| CN106980857A (en) * | 2017-02-24 | 2017-07-25 | 浙江工业大学 | A kind of Brush calligraphy segmentation recognition method based on rubbings |
| CN107742309A (en) * | 2017-10-20 | 2018-02-27 | 杭州电子科技大学 | The axis computational methods of simple polygon based on LR sequences |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113297897A (en) | 2021-08-24 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108805957B (en) | Vector diagram generation method and system based on bitmap image self-adaptive segmentation | |
| CN112597876B (en) | Handwriting Chinese character judging method based on feature fusion | |
| CN102968764B (en) | Chinese character image inpainting method based on strokes | |
| CN111723585A (en) | Style-controllable image text real-time translation and conversion method | |
| CN111323037B (en) | Voronoi path planning algorithm for novel framework extraction of mobile robot | |
| CN112712273B (en) | Handwriting Chinese character aesthetic degree judging method based on skeleton similarity | |
| CN107292936B (en) | Chinese character font vectorization method | |
| CN104992143B (en) | A kind of Chinese-character stroke extraction method towards vector font | |
| CN102332168B (en) | V (Voronoi) diagram-based raster map vectorization method | |
| CN111626297A (en) | Character writing quality evaluation method and device, electronic equipment and recording medium | |
| CN104156721A (en) | Off-line Chinese character stroke extraction method based on template matching | |
| CN112784531B (en) | Chinese character form and character library generation method based on deep learning and component splicing | |
| CN101393693B (en) | Computer educating method for Chinese character writing | |
| CN111523622B (en) | Method for simulating handwriting by mechanical arm based on characteristic image self-learning | |
| CN109409211B (en) | Processing method, processing device and storage medium for Chinese character skeleton stroke segments | |
| CN104951788A (en) | Extracting method of strokes of separate character in calligraphy work | |
| CN116740750A (en) | Reconstruction method and device for identifying size information based on engineering drawing | |
| CN113297897B (en) | Writing brush character auxiliary exercise method based on LR sequence and center axis information | |
| CN115116076A (en) | Character recognition method based on stroke extraction | |
| Zhou et al. | A novel text line segmentation method based on contour curve tracking for Tibetan historical documents | |
| WO2023065397A1 (en) | Stroke order recognition method and system for handwritten chinese character image | |
| CN112270704B (en) | Star point centroid extraction method based on data linked list structure | |
| CN115841671B (en) | Handwriting skeleton correction method, system and storage medium | |
| Boraik et al. | Characters segmentation from Arabic handwritten document images: hybrid approach | |
| CN115984875B (en) | Stroke similarity evaluation method and system for hard-tipped pen regular script copy work |
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 |