Scribble matching
First Claim
1. A method of matching samples of electronically stored freehand input with other stored samples of freehand input, the method comprising:
- locating velocity minima in each said freehand input sample;
coding each said freehand input sample by a sequence of symbols each representing a feature of the freehand input sample at a velocity minimum which is one of said velocity minimum; and
matching each said sequence of symbols against codes of said other stored samples of freehand input using a string edit distance metric;
wherein at least one of said freehand input samples is a figure exceeding a single text character, and said matching is completed for each said freehand input sample without identification of said freehand input sample as a sequence of individual text characters.
2 Assignments
0 Petitions
Accused Products
Abstract
The present invention relates to scribble matching methods, ie. methods for searching samples of electronic ink without having first to translate the ink into ASCII text. The methods of the present invention find application in pen-based computer devices. The invention entails the identification and encoding of velocity minima in the input scribble. Three scribble matching algorithms are described--a syntactic matcher, a word matcher and an elastic matcher. The syntactic matcher encodes the input scribble according to the shape of the scribble in the region of the velocity minima. The word matcher encodes the input scribble according to the heights of the velocity minima from a reference line, using a continuous height encoding scheme. The elastic matcher encodes the input scribble according to a classification of the velocity minima and intermediate points. The use of velocity minima in the encoding of scribble input provides a stable set of points and leads to improved performance in comparison with known scribble matching methods.
-
Citations
29 Claims
-
1. A method of matching samples of electronically stored freehand input with other stored samples of freehand input, the method comprising:
-
locating velocity minima in each said freehand input sample; coding each said freehand input sample by a sequence of symbols each representing a feature of the freehand input sample at a velocity minimum which is one of said velocity minimum; and matching each said sequence of symbols against codes of said other stored samples of freehand input using a string edit distance metric; wherein at least one of said freehand input samples is a figure exceeding a single text character, and said matching is completed for each said freehand input sample without identification of said freehand input sample as a sequence of individual text characters. - View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 20)
-
-
19. An apparatus for matching samples of electronically stored freehand input with other stored samples of freehand input, said apparatus comprising:
-
a velmin locator for locating velocity minima in each said freehand input sample; a freehand input encoder responsive to velocity minima received from said velmin locator, said freehand input encoder coding each said freehand input sample by a sequence of symbols, each symbol representing a feature of the freehand input at a velocity minimum which is one of said velocity minima; and a matcher responsive to said sequence of symbols received from said encoder, said matcher matching the sequence of symbols against codes of other samples of freehand input by using a string edit distance metric; wherein at least one of said freehand input samples is a figure exceeding a single text character, and said matcher matches each said freehand input sample without identifying it as a sequence of individual text characters. - View Dependent Claims (25)
-
-
21. A method of matching samples of electronically stored freehand input, said method comprising:
-
locating velocity minima in a first freehand input sample; coding the first freehand input sample by a sequence of symbols each representing a feature of the first freehand input sample at a velocity minimum which is one of said velocity minima; storing said coded first freehand input sample; repeating said steps of locating velocity minima, coding and storing for each of a plurality of further first freehand input samples to obtain a stored set of coded first freehand input samples; locating velocity minima in a second freehand input sample; coding the second freehand input sample by a sequence of symbols each representing a feature of the second freehand input sample at a velocity minimum; and matching the sequence of symbols from the second freehand input sample against the stored set of coded first freehand input samples using a string edit distance metric; wherein at least one of said first freehand input samples is a figure exceeding a single character, and said matching is completed without identification of said at least one of said first freehand input samples sequence of individual text characters. - View Dependent Claims (22, 23, 24)
-
-
26. A method of matching a test sample of electronically stored freehand input with other stored samples of freehand input, said method comprising:
-
locating velocity minima in the test sample of freehand input; coding the test sample of freehand input according to a plurality of coding steps, wherein each coding step comprises coding the test sample of freehand input by a sequence of symbols each representing a feature of the test sample of freehand input at a velocity minimum which is one of said velocity minima; for each sequence of symbols generated by one of said plurality of coding steps, matching the sequence of symbols against codes of other samples of freehand input using a string edit distance metric to provide a match score; and calculating a weighted sum of the match scores to obtain an overall match score for each of the other samples of freehand input; wherein the method of matching is performed on a plurality of test samples at least one of which is a figure exceeding a single text character, and matching does not include identification of any of said plurality of test samples as a string of individual text characters. - View Dependent Claims (27, 28, 29)
-
Specification