國立高雄大學圖資館 |

Language: English

Back

Discerning structure from freeform sketches

Record Type:	Electronic resources : Monograph/item
Title/Author:	Discerning structure from freeform sketches
Author:	Shilman, Michael M.
Description:	183 p.
Notes:	Chair: A. Richard Newton.
Notes:	Source: Dissertation Abstracts International, Volume: 65-02, Section: B, page: 0849.
Contained By:	Dissertation Abstracts International65-02B.
Subject:	Computer Science.
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3121695
ISBN:	0496690469

Discerning structure from freeform sketches
Shilman, Michael M.

Discerning structure from freeform sketches [electronic resource] - 183 p.

Chair: A. Richard Newton.

Thesis (Ph.D.)--University of California, Berkeley, 2003.

Each of these techniques is built on a general recognition architecture that uses a MultiTree protocol to compactly represent multiple tree hypotheses. The protocol enables functional and concurrent composition of off-the-shelf recognizers. The techniques are evaluated using different tree edit distance metrics.

ISBN: 0496690469Subjects--Topical Terms:

212513
Computer Science.

Discerning structure from freeform sketches
LDR:03468nmm _2200337 _450 001 162441
005 20051017073447.5
008 230606s2003 eng d
020 $a 0496690469
035 $a 00148942
035 $a 162441
040 $a UnM $c UnM
100 0 $a Shilman, Michael M. $3 227580
245 1 0 $a Discerning structure from freeform sketches $h [electronic resource]
300 $a 183 p.
500 $a Chair: A. Richard Newton.
500 $a Source: Dissertation Abstracts International, Volume: 65-02, Section: B, page: 0849.
502 $a Thesis (Ph.D.)--University of California, Berkeley, 2003.
520 # $a Each of these techniques is built on a general recognition architecture that uses a MultiTree protocol to compactly represent multiple tree hypotheses. The protocol enables functional and concurrent composition of off-the-shelf recognizers. The techniques are evaluated using different tree edit distance metrics.
520 # $a For document annotations, I present a style-independent recognition algorithm for common annotation markup structures. The algorithm combines minimal temporal assumptions with spatial statistics to yield accurate results. Among other things, the results of this recognition can be used to reflow annotation markup as the underlying document layout changes.
520 # $a For note-taking, I present an algorithm to distinguish handwriting from drawings in arbitrary, freeform notes. The algorithm is language-independent, depends only minimally on writing order, and detects writing at any angle on the page.
520 # $a For sketched diagrams, I present a general recognition framework based on context-free visual language grammars and statistical language parsing. The approach is evaluated on a language for user interface design and generalizes to a variety of inking scenarios.
520 # $a These claims are validated by considering three separate scenarios in detail: note-taking in meetings, annotating digital documents with markup, and sketching diagrams. For each of these common tasks, recognition should be flexible enough to incorporate the writing and drawing styles of an individual user, yet accurate enough to satisfy the scenarios for which it was intended. To this end, I have developed several algorithms that attempt to satisfy these constraints.
520 # $a This work explores user interfaces and algorithms for recognizing freeform sketching in digital ink. Like other recognition work before it, it is based on the premise that an increased understanding of the ink results in increased utility to the user. Interpreted digital ink is preferable to physical ink for its archival, search, editing, and simulation capabilities and its relative ease of integration with existing computer applications. However, unlike previous recognition work it also posits that a computer should never obstruct the user's thought capture in creative scenarios. In other words, the primary goal of a pen computer should be to unobtrusively capture freeform digital ink; the user should not be concerned with how the ink will be interpreted.
590 $a School code: 0028.
650 # 0 $a Computer Science. $3 212513
650 # 0 $a Engineering, Electronics and Electrical. $3 226981
690 $a 0544
690 $a 0984
710 0 # $a University of California, Berkeley. $3 212474
773 0 # $g 65-02B. $t Dissertation Abstracts International
790 $a 0028
790 1 0 $a Newton, A. Richard, $e advisor
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://libsw.nuk.edu.tw/login?url=http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3121695 $z http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3121695