The first examples of handwriting recognition occurred in the 1980s with the Pencept Penpad and the Inforite point-of-sale terminal. The first sale of such products was in the later 80s. Handwriting recognition became more of a public concern with the release of Apple’s Newton. In the Newton, handwriting recognition was first used as an input method for a PDA. One of the larger subsequent handwriting recognition jumps came with the Palm OS’s Graffiti handwriting system. This system of a program-defined handwriting was the basis for our system. Some still remember Graffiti today, but handwriting recognition based on computer input now has a new frontrunner: Windows 7. Windows 7 handwriting recognition contains new advances and is surprisingly accurate. Handwriting recognition like OCR continues to be extremely useful to some PDA users and postoffices today.
As far as our project goes, we are mostly interested in the process of recognizing characters while the computer already has data about what the character looks like. This is actually not always the case as with systems of OCR that must first find characters and go through a process to get the initial data in a system like that of Graffiti.
There are actually many different ways of recognizing characters. Although our work on this subject is not yet extensive, we plan to implement our handwriting recognition through a few simple mathematical systems. Essentially our method will often scan through the written character for the number of intersections with a given axis or equation. Other methods of developing this tree of characters will include the location of the end of a given stroke as opposed to the beginning of that stroke. Methods of character recognition often include much more advanced math that we may need to implement if our tree does not branch to all possible characters.
Experts in the field of handwriting recognition seem to be more commonly commercial than academic. In some instances handwriting recognition can be viewed as more of a gimmick than anything else, but technology in the field has been advancing for over thirty years with the help of large companies.
5. Our current project sources include:
- Online stroke modeling for handwriting recognition
- http://portal.acm.org/citation.cfm?id=1463796
- Polynomial Approximation in Handwriting Recognition
- http://www.csd.uwo.ca/~watt/pub/reprints/2011-snc-polyhwr.pdf
- Background for Signature Verification
- http://www.lems.brown.edu/~dfs/presentation.htm
- General Information about Handwriting
- http://en.wikipedia.org/wiki/Handwriting_recognition
- Lengthy History of Computer Related Gesture Recognition
- http://users.erols.com/rwservices/pens/biblio70.html#Harmon62
- GUI Example of Modern Application
- http://www.the-graphics-tablet.com/handwriting-recognition.html
We have used further sources that will be included in a final bibliography, but these have been our initial sources of information.
6. Implementation of our program will be in the form of a simple program that will most likely implement a simple decision tree-based letter recognizing algorithm. Although we have very little time now to make this, I still am under the impression that this will be possible as I have designed a portion of this decision tree on paper.
Our Current Work Plan
Saturday: 6 – ? hours of straight programming.
Sunday: 8 – ? hours of straight programming.
Our presentation will need to include various historical information in order to live up to the required thirty minute length. There is a large amount of information on the topic and a portion of the presentation will include other methods of handwriting recognition.