features - Features Extraction for SketchBased Recognition...

Info iconThis preview shows pages 1–5. Sign up to view the full content.

View Full Document Right Arrow Icon
1 Features Extraction for Sketch- Based Recognition Lecture #9: Feature Extraction Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr. Joseph J. LaViola Jr. Fall 2009 Recall Pen-Based Interface Dataflow Raw Stroke Preprocessing Segmentation Data Feature Extraction And Classification Ink Parsing Sketch Understanding Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr. Analysis Make Inferences
Background image of page 1

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
2 Feature Extraction and Analysis ± What came first, the feature or the machine learning algorithm? ± Want to distinguish sketch components from one another ± Good features are critical ± Extract important information ² geometrical, statistical, contextual Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr. ± Examples include ² arc length, histograms, cusps, aspect ratio ² self-intersections, stroke area, etc… Finding Features ± Challenging problem ² need fast algorithms for gathering information ² features must be good discriminators ± Often trial and error ± Can be domain specific Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr.
Background image of page 2
3 Geometric Features (1) ± Number of strokes ² if you know how many strokes a symbol has, you can break up your recognizer into pieces (i e can break up your recognizer into pieces (i.e., recognizer for 1 stroke symbols, recognizer for 2 stroke symbols …) ± Cusps ² smooth vs. jagged strokes distance between cusps Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr. ² ± useful for when cusps are close together/far apart Geometric Features (2) ± Aspect ratio (width / height) ² tall vs. flat ± Self Intersections ² loops vs. no loops ² strokes with write over ² distance between self intersections also useful ² use line segment intersection algorithm Fall 2009 CAP 6105 – Pen-Based User Interfaces ©Joseph J. LaViola Jr. loops write over
Background image of page 3

Info iconThis preview has intentionally blurred sections. Sign up to view the full version.

View Full DocumentRight Arrow Icon
4 Geometric Features (3) ± First and last distance ² Strokes where first and last points are close together vs far apar together vs. far apart ² simple computation – ± Arc length ²
Background image of page 4
Image of page 5
This is the end of the preview. Sign up to access the rest of the document.

This note was uploaded on 06/12/2011 for the course CAP 6105 taught by Professor Lavoila during the Spring '09 term at University of Central Florida.

Page1 / 15

features - Features Extraction for SketchBased Recognition...

This preview shows document pages 1 - 5. Sign up to view the full document.

View Full Document Right Arrow Icon
Ask a homework question - tutors are online