handwriting
所属分类:模式识别(视觉/语音等)
开发工具:Visual C++
文件大小:3756KB
下载次数:2
上传日期:2016-06-05 16:25:45
上 传 者:
续写@那段情
说明: 手写识别系统,可识别手写文字,并转化为标准字符库中文字
(handwriting recognize)
文件列表:
handwriting\INSTALL (1109, 2003-03-11)
handwriting\Makefile (10493, 2003-06-05)
handwriting\data (0, 2003-06-05)
handwriting\data\FeatureMatrix (2285, 2003-01-28)
handwriting\data\_horizontal-line.0 (68671, 2003-01-28)
handwriting\data\default (0, 2003-06-05)
handwriting\data\default\greek (0, 2003-06-05)
handwriting\data\default\greek\lower-case (0, 2003-06-05)
handwriting\data\default\greek\lower-case\FeatureMatrix (42821, 2003-01-28)
handwriting\data\default\greek\lower-case\_alpha.0 (16932, 2003-01-28)
handwriting\data\default\greek\lower-case\_beta.0 (43763, 2003-01-28)
handwriting\data\default\greek\lower-case\_delta.0 (28837, 2003-01-28)
handwriting\data\default\greek\lower-case\_epsilon.0 (23085, 2003-01-28)
handwriting\data\default\greek\lower-case\_eta.0 (35618, 2003-01-28)
handwriting\data\default\greek\lower-case\_gamma.0 (19086, 2003-01-28)
handwriting\data\default\greek\lower-case\_lambda.0 (28974, 2003-01-28)
handwriting\data\default\greek\lower-case\_mu.0 (26139, 2003-01-28)
handwriting\data\default\greek\lower-case\_nu.0 (27294, 2003-01-28)
handwriting\data\default\greek\lower-case\_omega.0 (24956, 2003-01-28)
handwriting\data\default\greek\lower-case\_phi.0 (40480, 2003-01-28)
handwriting\data\default\greek\lower-case\_pi.0 (29992, 2003-01-28)
handwriting\data\default\greek\lower-case\_psi.0 (48407, 2003-01-28)
handwriting\data\default\greek\lower-case\_rho.0 (25814, 2003-01-28)
handwriting\data\default\greek\lower-case\_sigma.0 (33458, 2003-01-28)
handwriting\data\default\greek\lower-case\_tau.0 (31904, 2003-01-28)
handwriting\data\default\greek\lower-case\_theta.0 (28847, 2003-01-28)
handwriting\data\default\greek\lower-case\_xi.0 (45793, 2003-01-28)
handwriting\data\default\greek\lower-case\_zeta.0 (52643, 2003-01-28)
handwriting\data\default\greek\upper-case (0, 2003-06-05)
handwriting\data\default\greek\upper-case\Delta.0 (36293, 2003-01-28)
handwriting\data\default\greek\upper-case\FeatureMatrix (24806, 2003-01-28)
handwriting\data\default\greek\upper-case\Gamma.0 (27727, 2003-01-28)
handwriting\data\default\greek\upper-case\Lambda.0 (38795, 2003-01-28)
handwriting\data\default\greek\upper-case\Omega.0 (29297, 2003-01-28)
handwriting\data\default\greek\upper-case\Phi.0 (47397, 2003-01-28)
handwriting\data\default\greek\upper-case\Pi.0 (29186, 2003-01-28)
handwriting\data\default\greek\upper-case\Psi.0 (61757, 2003-01-28)
handwriting\data\default\greek\upper-case\Sigma.0 (34495, 2003-01-28)
handwriting\data\default\greek\upper-case\Theta.0 (46896, 2003-01-28)
... ...
==========================
ABOUT JIMHR
==========================
JIMHR or "Java Interactive Mathematical Handwriting Recognizer" as the name
suggests is the online handwriting recogition system that specializes in the
mathematical domain. It processes a user's handwriting through a mouse or
stylus pen and outputs the corresponding mathematical formula in real-time in
a type set image and either Latex or MathML. With the help of a sophisticated
graphical user interface, the program also allows a user to correct any
mistakes resulting from misclassification or to rearrange the symbols to suit
his needs. Other operations include delete, undo, and redo. JIMHR runs on
multiple platforms, including UNIX, Linux, and Windows.
JIMHR is based on The Freehand Formula System Entry System and Diagram
Recognition Application for Computer Understanding of Large Algebraic
Expressions (FFES/DRACULAE) project at Queens University in Ontario, Canada
(FFES originated at the University of Otago, New Zealand) and The Natural Log
System (NLS) project at MIT. The goal of JIMHR is to combine the algorithmic
efficiencies and functionalities from both projects to promote further
collaborative research and development in the field, while also providing a
viable, open-source solution for mathematical handwriting recognition (MHR)
that can be used and stress-tested in commercial and educational environments.
For more details regarding these two projects, you are advised to visit their
web sites at:
http://www.cs.queensu.ca/drl/ffes/
http://www.ai.mit.edu/projects/lv/projects/notebook/
Two research papers for reference are included in the 'doc' directory.
JIMHR consists of four main components: character recognizer, expression
partitioner, parser, and graphical user interface. The character
recognizer and expression partitioner are based on Caltech Interface Tools
(CIT) library written in C++ by James Arvo, et al. (for details see
http://www.cs.caltech.edu/~arvo/software.html). While the graphical
user interface is written in Java, and the parser is a command line tool called
DRACULAE which was originally developped at Queen's University by Richard
Zanibbi, under the supervision of Dorothea Blostein and James R. Cordy. It is
implemented the TXL programming language (for more information regarding TXL,
please checkout website http://www.txl.ca).
JIMHR is released under LGPL license.
The project homepage is at http://jimhr.sourceforge.net
==========================
CVS REPOSITORY
==========================
The latest jimhr source code can always be founded by anonymnous
CVS at cvs.jimhr.sourceforge.net
If you have questions, comments, suggestions or would like to join
the development team please email to joygong@users.sourceforge.net
==========================
PROGRAM STRUCTURE
==========================
Jimhr contains 7 main directories. They are listed with the explanation below.
data/
contains a training set for all math characters that are used by the
character recognizer.
doc/
contains pdf files that describe algorithms and mechanisms behind FFES and NLS
program
include/
contains C++ header files used in the character recognizer, expression
partitioner, and JNI header files
jimhr/
contains java files which are used to implement graphical user interface and
control the overall flow of jimhr
lib/
contains jar and zip files used in displaying LaTeX expression on Java Canvas
shell/
contains executable jimhr programs for both Unix and Window system. It also
contains command line tools GetTeX which is used as a parser
src/
contains C++ source files used in the character recognizer and, expression
partitioner, and JNI source files
==========================
CODE DETAILS
==========================
As mentioned before jimhr consists of four main components, ie. character
recognizer, expression partitioner, parser, and graphical user interface.
This section is an attempt to explain the interactions between these modules
and pertinent source codes.
The diagram below shows the overall structure of jimhr
GUI
*************************************
* *
* *
* *
* *
* *
* Drawing Area *
* *
--------- * *
| * *
| * *
| * *
| *************************************
| * * *
| * MathML/ Latex * Rendered *
| * Output * Latex *
| * * Output *
| *************************************
| ^ ^
| /|\ /|\
| | |
| |------------------|-------------------------|
| |
\|/ |
v |
**************** ************* ************ |
* Expression * * Symbol * * * |
* *--------->* *---------->* Parser *---------
* Partiitioner * * Recognizer* * *
**************** ************* ************
GUI (jimhr/gui/Demo.java) has three display areas. The top one
(jimhr/gui/InkCanvas.java) accepts handwriting input from a user and record
all the coordinates of each stroke. It then send the data to the expression
partitioner (jimhr/recognizer/Segmenter.java) which is responsible for
partitioning stokes into different groups where each group corresponding to a
symbol. Each group of strokes in turn is sent out to the symbol recognizer
which produces a sequence of symbols. These symbols are then assembled into a
LaTeX expression by the parser (shell/GetTeX command line tool). The bottom
left panel of GUI displays the LaTeX expression, while the one in the bottom
right (jimhr/gui/ExpressionCanvas.java) renders it into typical mathematical
expression.
==========================
CONTRIBUTORS
==========================
JoyGong Ho started the project and
made the initial working implementation.
The JIMHR project team would like to thank the contributors for all of their
work.
CIT
===
James Arvo
Xavier Fan
Vladimir Fedorov
Yuanshan Guo
Kevin Novins
Steve Smithies
FFES/DRACULAE
=============
Steve Smithies
Kevin Novins
Richard Zanibbi
Arlis Rose
David Tausky
Nick Willan
Eric Bailey
NLS
===
Paul Viola
Nicholas Matsakis
Erik Miller
Jeremy De Bonet
近期下载者:
相关文件:
收藏者: