CN107122768A - A kind of three-dimensional pen type identification preprocess method - Google Patents
A kind of three-dimensional pen type identification preprocess method Download PDFInfo
- Publication number
- CN107122768A CN107122768A CN201710402133.5A CN201710402133A CN107122768A CN 107122768 A CN107122768 A CN 107122768A CN 201710402133 A CN201710402133 A CN 201710402133A CN 107122768 A CN107122768 A CN 107122768A
- Authority
- CN
- China
- Prior art keywords
- mrow
- msub
- munderover
- sample point
- matrix
- 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.)
- Pending
Links
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
-
- 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
- G06V30/22—Character recognition characterised by the type of writing
- G06V30/228—Character recognition characterised by the type of writing of three-dimensional handwriting, e.g. writing in the air
Landscapes
- Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Processing Or Creating Images (AREA)
Abstract
The present invention relates to a kind of three-dimensional pen type identification preprocess method, belong to pattern-recognition and artificial intelligence field.Comprise the following steps：Obtain not in the same time under a three-dimensional coordinate；Find k Neighbor Points of each three-dimensional coordinate；The partial reconstruction weight matrix of the sample point is calculated by the Neighbor Points of each sample point；The output matrix of the sample point is calculated by the partial reconstruction weight matrix and its Neighbor Points of the sample point；Projection output is finally determined according to the output matrix of each sample point.Three-dimensional pen type track is carried out dimensionality reduction degree processing by the present invention by finding best projection plane, the content transformation that rapidly and effectively can be inputted user under three dimensions is the content on two dimensional surface, the problem of writing of three dimensions pen type is not easy to identify is effectively solved, the accuracy of pen-based interaction, fluency and high efficiency under three dimensions is realized.
Description
Technical field
The invention belongs to pattern-recognition and artificial intelligence field, more particularly to a kind of three-dimensional space based on ultrasonic sensor
Between handwriting recognition preprocess method.
Background technology
With continuing to develop for hardware and software technology, the natural man-machine interaction of human beings is already possibly realized,
The continuous proposition of emerging interactive conceptual and interaction technique, allows man-machine interaction mode to be increasingly becoming study hotspot.Wherein pen type
The traditional use pen custom of people has been continued to use in interaction, can be reached the easy interaction effect of nature, how be handled under three dimensions
Pen type track has very important significance to three-dimensional pen-based interaction.
The more than ten years have been developed in handwriting recognition technology, can be divided into plane handwriting recognition technology according to the difference in writing space
With space handwriting recognition technology.If the space handwritten content to user is identified, user only needs to hold embedded ultrasonic wave
The pen-based interaction equipment of sensor is freely write in the air, and it is neither limited (such as handwriting pad, touch by some specific plane
Screen), it is not required that the limitation of any other auxiliary equipment (such as camera), this will be a kind of brand-new man-machine interaction realization side
Formula.Virtual reality, remote interaction technology popularization simultaneously, space handwriting recognition will have more application scenarios and development is empty
Between.
It is plane handwriting recognition problem for space handwriting recognition is returned, 2011, Tan Xiaofeng, Shen Haibin etc. was proposed
, planarization process has been carried out to space hand-written character track using the projection algorithm based on pivot analysis.At present for space
The research of handwriting recognition technology is also relatively fewer, the processing side not generally acknowledged also to the flow for how handling space hand-written character
Method.No matter writer wants to carry out the input of content or carries out man-machine interaction, and how effective by computer the content of input is
Identification, how computer further feeds back the intention of writer, and there is presently no a kind of effective method.If utilization space hand
The characteristics of write characters, planarizes space handwriting, and processing identification feasibility is carried out using existing plane handwriting recognition technology
Height, so how adaptively to find optimal projection plane is the key for solving this problem.
The content of the invention
The present invention provides a kind of three-dimensional pen type identification preprocess method, and to solve, current user's space is hand-written to be can not find most preferably
Projection plane, can not further the carry out three dimensions pen-based interaction of efficient natural the problem of.
The present invention is adopted the technical scheme that, is comprised the following steps：
(1), obtain not in the same time under three-dimensional pen type trajectory coordinatesWill each three-dimensional pen type rail
Mark coordinate is considered as sample point, has n sample point；
(2), by searching out k Neighbor Points Z of each sample point apart from parameterij, wherein Zij(j=1,2 ... k) be Xi
K Neighbor Points,
(3) the partial reconstruction weights of each sample point, are calculated, the partial reconstruction weight matrix W of sample point is drawn, draws
Enter error function min ε (W) and partial reconstruction weight matrix W, computational methods such as formula (1) institute are obtained by method of Lagrange multipliers
Show：
Wherein min ε (W) are the error function of definition,It is XiWith ZijBetween weights, and to meet：
(4) output matrix S, is calculated according to partial reconstruction weight matrix W, loss function value min ε (S) is introduced and utilizes glug
Bright day multiplier method obtains optimal solution under constraints, shown in computational methods such as formula (2)：
Wherein, min ε (Y) are loss function value, SiFor XiOutput vector, Qij(j=1,2 ... k) be SiK neighbour
Point, and to meet：
Wherein, I is d*d unit matrix, and d is output matrix dimension, willIt is stored in matrix W, works as ZijIt is XiIt is near
During adjoint point,Otherwise, WijFor 0, wherein WijIt is the value of the i-th row jth row in W matrixes, then formula (2) can be rewritten as：
Wherein, M=(I-W) (I-W)T, MijIt is the value of the i-th row jth row in Metzler matrix；
(5) characteristic value and characteristic vector of Metzler matrix, are solved, characteristic value is arranged from small to large, casts out first feature
Value, take the 2nd to the corresponding characteristic vector of (d+1) individual characteristic value as output result.
In step (2) of the present invention is apart from expressed as parameters：
Generally p=2, diqRepresent sample
Point XiWith sample point XiqEuclidean distance.
Each sample point is obtained in step (2) of the present invention apart from the Euclidean distance of all sample points by from small to large
Sequence, takes the preceding 60% corresponding sample point of Euclidean distance as Neighbor Points, determines k values.
The matrix dimension d of step (4) output matrix of the present invention is 2.
It is adaptive by three dimensional non-linear data the method have the advantages that the coordinate of three-dimensional pen type track carried out into dimension-reduction treatment
It is mapped in two-dimensional space, still is able to keep the topological relation of legacy data after processing, this process is by space hand-written character rail
Mark is planarized, and it is high to carry out processing feasibility using existing plane handwriting recognition technology, it is possible to achieve the action of space pen type, pen type
Effective identification of input content, completes the pen-based interaction under three dimensions, and the content for pen type under three dimensions is inputted still
Man-machine interaction has important meaning.
Brief description of the drawings
Fig. 1 is the flow chart of the present invention；
Fig. 2 is three-dimensional hand-written character artwork of the invention；
Fig. 3 is the pretreated design sketch of three-dimensional hand-written character of the invention.
Embodiment
The input equipment implemented used in the present invention is the three-dimensional pen-based interaction device based on ultrasonic sensor, based on ultrasound
The system flow chart of the three-dimensional pen type identification preprocess method of wave sensor is as shown in figure 1, comprise the following steps that：
(1), obtain not in the same time under three-dimensional pen type trajectory coordinatesWill each three-dimensional pen type rail
Mark coordinate is considered as sample point, has n sample point；
(2), by searching out k Neighbor Points Z of each sample point apart from parameterij, wherein Zij(j=1,2 ... k) be Xi
K Neighbor Points,
(3) the partial reconstruction weights of each sample point, are calculated, the partial reconstruction weight matrix W of sample point is drawn, draws
Enter error function min ε (W) and partial reconstruction weight matrix W, computational methods such as formula (1) institute are obtained by method of Lagrange multipliers
Show：
Wherein min ε (W) are the error function of definition,It is XiWith ZijBetween weights, and to meet：
(4) output matrix S, is calculated according to partial reconstruction weight matrix W, loss function value min ε (S) is introduced and utilizes glug
Bright day multiplier method obtains optimal solution under constraints, shown in computational methods such as formula (2)：
Wherein, min ε (Y) are loss function value, SiFor XiOutput vector, Qij(j=1,2 ... k) be SiK neighbour
Point, and to meet：
Wherein, I is d*d unit matrix, and d is output matrix dimension, willIt is stored in matrix W, works as ZijIt is XiIt is near
During adjoint point,Otherwise, WijFor 0, wherein WijIt is the value of the i-th row jth row in W matrixes.Then formula 2 can be rewritten as：
Wherein, M=(I-W) (I-W)T, MijIt is the value of the i-th row jth row in Metzler matrix；
(5) characteristic value and characteristic vector of Metzler matrix, are solved, characteristic value is arranged from small to large, casts out first feature
Value, take the 2nd to the corresponding characteristic vector of (d+1) individual characteristic value as output result.
In step (2) of the present invention is apart from expressed as parameters：
Generally p=2, diqRepresent sample
Point XiWith sample point XiqEuclidean distance；
Each sample point is obtained in step (2) of the present invention apart from the Euclidean distance of all sample points by from small to large
Sequence, takes the preceding 60% corresponding sample point of Euclidean distance as Neighbor Points, determines k values；
The matrix dimension d of step (4) output matrix of the present invention is；
The present invention is further illustrated by instantiation below.
The main difference of space hand-written character and plane hand-written character is that space hand-written character does not only have length and width
Degree, also with this attribute of thickness.If it is improper that projection plane is chosen, it is certain to greatly differ from each other with the intention of writer.This reality
Three-dimensional lettering pen and supporting acquisition software of the example based on ultrasonic sensor, acquire the data of 9 experimenters, each experimenter
The three-dimensional lettering pen based on ultrasonic sensor is held in three dimensions handwritten numeral characters 0 to 9, each experimenter is in three-dimensional space
Between in each numerical character of writing 5 times, gathered data 450 times, obtains each numerical character size in space, such as following table institute altogether
Show：
Be concluded that, space numerical character write with plane writing style according to writer, except character " 1 " it
Outside, the average value that the average value of the length and width of character should be much than thickness is big, and handwriting tracks are substantially curved
Form.A projection plane can be so determined, as long as finding character trace most flat observed direction and perpendicular to this side
To projection, it is possible to obtain optimal drop shadow effect, and for character " 1 ", the character area projected to length and thickness
Not less, so equally applicable above-mentioned projection principle.And when solving the characteristic value and characteristic vector of Metzler matrix, according to from it is small to
Big sequence first remains the corresponding characteristic vector of preceding 3 characteristic values, matrix is represented to turn into each vector in each projection
Projected length above vector, its characteristic value is then weight.The corresponding characteristic vector of characteristic value of minimum has been given up, throwing is remained
Those maximum components of shadow energy, so maximize the information for remaining matrix representative, while having given up the minimum of character trace
Size, obtains the most conceivable plane of writer.
Determine after projection plane, the direction of character is not determined also.But for the length and width of space written character
Degree is difficult to judge that character is correctly oriented, and is determined so remaining 2 kinds of different results for writer.
It can see from Fig. 2 and Fig. 3, can be effective using three-dimensional hand-written character preprocess method proposed by the present invention
Best projection plane identification hand-written character profile is found out, has good adaptivity to the action of writer, allows space pen type to know
There is not good application prospect, finally play its due potentiality.
Claims (4)
1. a kind of three-dimensional pen type identification preprocess method, it is characterised in that comprise the following steps：
(1), obtain not in the same time under three-dimensional pen type trajectory coordinatesWill each three-dimensional pen type trajectory coordinates
It is considered as sample point, has n sample point；
(2), by searching out k Neighbor Points Z of each sample point apart from parameterij, wherein Zij(j=1,2...k) is XiK
Individual Neighbor Points,(i=1,2...n)；
(3) the partial reconstruction weights of each sample point, are calculated, the partial reconstruction weight matrix W of sample point is drawn, introduces and misses
Difference function min ε (W) obtain partial reconstruction weight matrix W by method of Lagrange multipliers, shown in computational methods such as formula (1)：
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&epsiv;</mi>
<mrow>
<mo>(</mo>
<mi>W</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mo>|</mo>
<msub>
<mi>X</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<munderover>
<mi>&Sigma;</mi>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mi>j</mi>
</msubsup>
<msub>
<mi>Z</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>1</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein min ε (W) are the error function of definition,It is XiWith ZijBetween weights, and to meet：
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mi>j</mi>
</msubsup>
<mo>=</mo>
<mn>1</mn>
</mrow>
(4) output matrix S, is calculated according to partial reconstruction weight matrix W, loss function value min ε (S) are introduced using Lagrange
Multiplier method obtains optimal solution under constraints, shown in computational methods such as formula (2)：
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&epsiv;</mi>
<mrow>
<mo>(</mo>
<mi>S</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<mo>|</mo>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
<mo>-</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>j</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>k</mi>
</munderover>
<msubsup>
<mi>w</mi>
<mi>i</mi>
<mi>j</mi>
</msubsup>
<msub>
<mi>Q</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<msup>
<mo>|</mo>
<mn>2</mn>
</msup>
<mo>-</mo>
<mo>-</mo>
<mo>-</mo>
<mrow>
<mo>(</mo>
<mn>2</mn>
<mo>)</mo>
</mrow>
</mrow>
Wherein, min ε (Y) are loss function value, SiFor XiOutput vector, Qij(j=1,2 ... k) be SiK Neighbor Points, and
Meet：
<mrow>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
<mo>=</mo>
<mn>0</mn>
</mrow>
<mrow>
<mfrac>
<mn>1</mn>
<mi>n</mi>
</mfrac>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
<msup>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
<mi>T</mi>
</msup>
<mo>=</mo>
<mi>I</mi>
</mrow>
Wherein, I is d*d unit matrix, and d is output matrix dimension, willIt is stored in matrix W, works as ZijIt is XiNeighbor Points
When,Otherwise, WijFor 0, wherein WijIt is the value of the i-th row jth row in W matrixes, then formula (2) can be rewritten as：
<mrow>
<mi>m</mi>
<mi>i</mi>
<mi>n</mi>
<mi>&epsiv;</mi>
<mrow>
<mo>(</mo>
<mi>Y</mi>
<mo>)</mo>
</mrow>
<mo>=</mo>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>i</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<munderover>
<mo>&Sigma;</mo>
<mrow>
<mi>q</mi>
<mo>=</mo>
<mn>1</mn>
</mrow>
<mi>n</mi>
</munderover>
<msub>
<mi>M</mi>
<mrow>
<mi>i</mi>
<mi>j</mi>
</mrow>
</msub>
<msup>
<msub>
<mi>S</mi>
<mi>i</mi>
</msub>
<mi>T</mi>
</msup>
<msub>
<mi>S</mi>
<mi>q</mi>
</msub>
<mo>,</mo>
</mrow>
Wherein, M=(I-W) (I-W)T, MijIt is the value of the i-th row jth row in Metzler matrix；
(5) characteristic value and characteristic vector of Metzler matrix, are solved, characteristic value is arranged from small to large, casts out first characteristic value, takes
2nd is used as output result to the corresponding characteristic vector of (d+1) individual characteristic value.
2. a kind of three-dimensional pen type identification preprocess method according to claim 1, it is characterised in that in the step (2)
Be apart from expressed as parameters：
Q=1,2..., n, generally p=2, diqRepresent sample point Xi
With sample point XiqEuclidean distance.
3. a kind of three-dimensional pen type identification preprocess method according to claim 1, it is characterised in that in the step (2)
Each sample point is obtained apart from the Euclidean distance of all sample points by sorting from small to large, takes preceding 60% Euclidean distance corresponding
Sample point determines k values as Neighbor Points.
4. a kind of three-dimensional pen type identification preprocess method according to claim 1, it is characterised in that the step (4) is defeated
The matrix dimension d for going out matrix is 2.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710402133.5A CN107122768A (en) | 2017-05-31 | 2017-05-31 | A kind of three-dimensional pen type identification preprocess method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201710402133.5A CN107122768A (en) | 2017-05-31 | 2017-05-31 | A kind of three-dimensional pen type identification preprocess method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN107122768A true CN107122768A (en) | 2017-09-01 |
Family
ID=59729141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201710402133.5A Pending CN107122768A (en) | 2017-05-31 | 2017-05-31 | A kind of three-dimensional pen type identification preprocess method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN107122768A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609593A (en) * | 2017-09-15 | 2018-01-19 | 杭州电子科技大学 | A kind of three dimensions hand-written character dimension reduction method based on most long track projection |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5799107A (en) * | 1993-05-31 | 1998-08-25 | Fujitsu Limited | Control system for pen-input type computer |
| US20080101702A1 (en) * | 2006-10-30 | 2008-05-01 | Fuji Xerox Co., Ltd. | Image generation apparatus, image processing apparatus, computer readable medium and computer data signal |
| CN102982349A (en) * | 2012-11-09 | 2013-03-20 | 深圳市捷顺科技实业股份有限公司 | Image recognition method and device |
| CN105139036A (en) * | 2015-06-19 | 2015-12-09 | 四川大学 | Handwritten figure identification method based on sparse coding |
| CN105740784A (en) * | 2016-01-25 | 2016-07-06 | 山东毅康科技股份有限公司 | Handwriting font identification system based on distance optimization dimension reduction |
-
2017
- 2017-05-31 CN CN201710402133.5A patent/CN107122768A/en active Pending
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5799107A (en) * | 1993-05-31 | 1998-08-25 | Fujitsu Limited | Control system for pen-input type computer |
| US20080101702A1 (en) * | 2006-10-30 | 2008-05-01 | Fuji Xerox Co., Ltd. | Image generation apparatus, image processing apparatus, computer readable medium and computer data signal |
| CN102982349A (en) * | 2012-11-09 | 2013-03-20 | 深圳市捷顺科技实业股份有限公司 | Image recognition method and device |
| CN105139036A (en) * | 2015-06-19 | 2015-12-09 | 四川大学 | Handwritten figure identification method based on sparse coding |
| CN105740784A (en) * | 2016-01-25 | 2016-07-06 | 山东毅康科技股份有限公司 | Handwriting font identification system based on distance optimization dimension reduction |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107609593A (en) * | 2017-09-15 | 2018-01-19 | 杭州电子科技大学 | A kind of three dimensions hand-written character dimension reduction method based on most long track projection |
| CN107609593B (en) * | 2017-09-15 | 2019-12-10 | 杭州电子科技大学 | Three-dimensional space handwritten character dimension reduction method based on longest track projection |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Chen et al. | Beyond human recognition: A CNN-based framework for handwritten character recognition | |
| Kara et al. | An image-based, trainable symbol recognizer for hand-drawn sketches | |
| CN103093196B (en) | Character interactive input and recognition method based on gestures | |
| Qu et al. | Data augmentation and directional feature maps extraction for in-air handwritten Chinese character recognition based on convolutional neural network | |
| CN105893968B (en) | The unrelated person's handwriting recognition methods end to end of text based on deep learning | |
| CN105844669A (en) | Video target real-time tracking method based on partial Hash features | |
| CN102622225A (en) | Multipoint touch application program development method supporting user defined gestures | |
| CN106886751A (en) | A kind of gesture identification method and system | |
| CN109033780A (en) | A kind of edge calculations access authentication method based on wavelet transformation and neural network | |
| CN101964053B (en) | On-line identification method of compound pattern | |
| Ye et al. | 3D sketch-based 3D model retrieval with convolutional neural network | |
| Shengtao et al. | Facial expression recognition based on global and local feature fusion with CNNs | |
| CN107220634B (en) | Based on the gesture identification method for improving D-P algorithm and multi-template matching | |
| Manal et al. | Survey on the approaches based geometric information for 3D face landmarks detection | |
| Wang et al. | Real-time visual static hand gesture recognition system and its FPGA-based hardware implementation | |
| CN103186241B (en) | A kind of interactive desktop contact right-hand man's recognition methods | |
| CN105678348B (en) | A kind of handwritten Chinese character normalization evaluation method and system | |
| CN104794741B (en) | The removing method and system of a kind of hand-drawing graphics puppet break | |
| Sooai et al. | Comparison of recognition accuracy on dynamic hand gesture using feature selection | |
| CN107122768A (en) | A kind of three-dimensional pen type identification preprocess method | |
| Wang et al. | Labanotation generation from motion capture data for protection of folk dance | |
| CN101604392B (en) | Chinese character stroke feature parallel extraction method used for on-line handwritten Chinese character recognition | |
| Lei et al. | Handwritten character recognition via direction string and nearest neighbor matching | |
| Chen | Research and application of chinese calligraphy character recognition algorithm based on image analysis | |
| CN101604378B (en) | Chinese character stroke feature serial extraction method used for on-line handwritten Chinese character recognition |
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 | ||
| RJ01 | Rejection of invention patent application after publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20170901 |