stw text detector

12
STW Text Detector Gili Werner

Upload: kirra

Post on 23-Feb-2016

52 views

Category:

Documents


0 download

DESCRIPTION

STW Text Detector. Gili Werner. Motivation. Detecting text in a natural scene is an important part of many Computer Vision tasks. Motivation. - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: STW Text Detector

STW Text DetectorGili Werner

Page 2: STW Text Detector

MotivationDetecting text in a natural scene is an

important part of many Computer Vision tasks

Page 3: STW Text Detector

MotivationFor example, the performance of optical

character recognition (OCR) algorithms can be highly improved by first identifying the regions of text in the image

Page 4: STW Text Detector

SWT Text DetectorIn this project I attempted to create a

powerful and reliable tool for detecting text regions in an image, by using the Stroke Width Transform (SWT)

grouping pixels together in an intelligent way, instead of looking for separating features of pixels

Page 5: STW Text Detector

The Stroke Width Transform3 major steps: 1. The stroke width transform

A stroke in the image is a continuous band of a nearly constant width

SWT is a local operator which calculates for each pixel the width of the most likely stroke containing the pixel

Page 6: STW Text Detector

The Stroke Width Transform2. Finding letter candidates

Grouping the pixels into letter candidates based on their stroke width

Page 7: STW Text Detector

The Stroke Width Transform3. Grouping letter candidates into regions of

text Group closely positioned letter candidates

into regions of text Filters out many falsely-identified letter

candidates, and improves the reliability of the algorithm results 

Page 8: STW Text Detector

Results

Page 9: STW Text Detector

Results

Page 10: STW Text Detector

StrengthsThe SW Detector can detect letters of

different languages (English, Hebrew, Arabic etc.)

The text can be of varying sizesThe text can be of different orientation

Including curvy textEven handwriting can be detected

Page 11: STW Text Detector

WeaknessesAppearance of noise

Foliage resembles lettersDoes not handle round and curved letters as

wellSmall and close letters tend to be grouped

together in the SW labeling phaseThese groups may be dismissed in the ‘finding

letter candidates’ phase

Page 12: STW Text Detector

Questions?