Image analysis – Pattern recognition – On-line recognition of handwritten characters
Reexamination Certificate
1997-09-08
2002-12-10
Chang, Jon (Department: 2623)
Image analysis
Pattern recognition
On-line recognition of handwritten characters
C382S311000
Reexamination Certificate
active
06493464
ABSTRACT:
BACKGROUND OF THE INVENTION
1. Field of the Invention
This invention relates to computer input systems in general, and more specifically to an apparatus and handwriting alphabet for use in a handwritten input and recognition system used in personal computing systems such as “palm-top” computers.
2. Description of Related Art
As computers have become increasingly popular for various applications, portable computers have been developed for a wide variety of uses. While many such portable computers use a traditional keyboard for input, for smaller computers, particularly including hand-held computers, the use of “pens” as an interface has been introduced as a way of making a small computer easier to use. With a pen interface, a user can place a pen or stylus directly on a touch-sensitive screen of the computer to control the software running on the computer and to input information: For many people, controlling a computer and entering text with a pen is more natural than using a keyboard.
An example of a prior art pen-based hand-held computer is shown in FIG.
1
A. The illustrated hand-held computer
1
is typically about 4 inches by 6.5 inches, with the majority of one surface comprising a touch-sensitive display screen
2
. The display screen
2
is typically a liquid crystal display (LCD) having a resolution of 256×320 pixels (although larger or smaller pixel arrays could be used). Various technologies can be used to sense the location of a pen or stylus
3
touched against the surface of the LCD screen
2
to indicate to the computer's operating system the X-Y coordinates of the touch. Various hardware buttons
4
may be provided to control different functions, and/or to turn power on or off to the unit. In addition, a variety of software buttons or icons
5
may be provided, in known fashion, to indicate such functions as, for example, word processing or a delete function. Computer-generated information is typically shown on the display screen
2
as ASCII characters
6
. One such hand-held computer is available as the “Zoomer” from Casio Corporation.
A common characteristic of such pen-based computers is the use of electronic “ink”. “Ink” comprises a series or trail of pixels changed (e.g., darkened or lightened) as a pen
3
is moved across the display screen
2
by a user, thus mimicking the application of real ink to paper.
Some prior art system designers suggest the use of unrecognized handwritten ink input. Although this approach works well for recording notes for personal use, it is not always suitable for data entry into a file which needs to be searched at a later date. In addition, ink requires considerably more storage space than ASCII characters. Accordingly, practical pen-based computers need a method of inputing text which usually includes some form of recognition system.
Various methods of recognizing handwriting are well known. One prior art approach is to provide a series of boxes in the input area (which is usually the display area) for entering character information. These systems use boxes for entry of text in an attempt to improve accuracy of recognition and to separate one character from the next. In these systems, an array of boxes is displayed and the user writes one character in each box. Although the boxes aid in improving the accuracy of recognition, most people find it awkward to write in boxes. Additionally, due to the number of boxes necessary to capture. even a short sentence, these systems are not very practical in a palm-top computer having a reduced data input area.
Another character recognition system is described in U.S. Pat. No. 5,125,039, entitled “Object Recognition System”, by the inventor of the present invention. In such a system, the user writes text without boxes in a free form manner. After a user inputs several ink characters, the computer applies special algorithms to separate the ink strokes into characters and then recognize each ink character as an ASCII character. It then replaces the ink representation of the characters drawn by the user with the standardized ASCII representation of those characters. Although these systems require less input area than boxed input systems, they are still difficult to implement on a palmtop computer having a small display. In addition, the computer has the additional burden of figuring out where one character ends and the next begins. This leads to recognition errors.
One additional major difficulty presented by prior art handwriting recognition systems is the delay time between text input and text recognition. The prior art systems typically require between 2 to 5 seconds after the user writes the ink character on the input tablet to recognize and display the ASCII character on a display device. In typical use, the prior art systems require the user to write a few words and then wait several seconds for the computer to start the recognition process. Alternatively, some systems (e.g., the “Newton” from Apple Computer) perform recognition without the user stopping and waiting. But in these systems, the words are still recognized several seconds after they are written. In all cases, the user cannot immediately realize when a recognition mistake has occurred. This type of handwritten text recognition system makes error correction difficult because the user must constantly look at the display for words which the user input several seconds before in order to be sure that text was correctly entered and correctly interpreted. Moreover, once a user detects an error, error correction is difficult because the user has to first select the word or characters which need to be corrected.
In summary, three of the major problems with current handwriting recognition systems are the delay from writing to recognition, the limited writing area of palmtop computers, and the difficulty of accurately recognizing separate characters in non-boxed entry systems.
Therefore, an improved pen data entry solution is needed which can accurately and efficiently recognize text on a small display. It has become evident that one crucial characteristic of such an improved solution is the ability to instantaneously (i.e., with little or no perceptible delay) recognize and display input text, similar to the response of currently available personal computers using keyboard input devices. Palm-top computers having the ability to instantly recognize and display text offer the user the opportunity to quickly recognize and correct mistakes. Instant recognition also permits the use of smaller input areas because the input area can be reused for writing subsequent characters.
One of the major impediments facing “instant” handwritten text recognition systems is presented by the multiple stroke (multi-stroke) characteristic of many English text characters. That is, many characters comprise more than one pen stroke. In this context, a single pen stroke is defined as a continuous movement of a pen while maintaining contact with a writing tablet. For example, the letters “T”, “H,” and “E” typically comprise multiple pen strokes, while the letters “S” and “O” typically comprise a single pen stroke. Prior art recognition systems have had difficulty in achieving essentially “instantaneous” recognition due to the fact that characters may comprise more than one pen stroke.
For example, due to the possibility that any given input character might be a multi-stroke character, it has been difficult to determine when a user has completed writing a one stroke (unistroke) character, or when the user is continuing to write a multi-stroke character. For example, a vertical line might represent the letter “I” or it could represent the first stroke in the multi-stroke letters “T”, “H” or “E”. In the past, recognition systems have solved this ambiguity by waiting until the user stopped writing, or by having a fixed delay. period after which characters were recognized, or by detecting the start of a next stroke sufficiently far from prior strokes as to indicate a new character. Each of these approaches are deficient due to the recognition time dela
Hawkins Jeffrey Charles
Marianetti, II Ron
Sipher Joseph Kahn
Chang Jon
Mahamedi Van
Palm Inc.
LandOfFree
Multiple pen stroke character set and handwriting... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Multiple pen stroke character set and handwriting..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Multiple pen stroke character set and handwriting... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2996613