Security and Communication Networks / 2017 / Article / Tab 3

Review Article

A Survey on Breaking Technique of Text-Based CAPTCHA

Table 3

Comparison of typical methods based on segmentation for breaking adherent CAPTCHA.

ExampleSourceSuccess rateReferenceBreaking methodYear

Google,
Yahoo
4.89%–66.2%[2]Segmentation: width
Recognition: CNN
2004

Microsoft
Google
Yahoo
61%
8.7%
25.9%
[4]Segmentation: color filling and projection
Recognition: CNN
2008

Hotmail40%[5]Segmentation: change width
Recognition: SVM
Post-processing: DP search
2009

MSN
Yahoo
18%
45%
[6]Segmentation: projection and central2010

Megaupload78%[36]Segmentation: color filling
Combination: nonredundancy
Recognition: CNN
2010

reCAPT-CHA Google33%
46.75%
[38]Segmentation: character structure feature
Recognition: CNN
2011

Yahoo54.7%[44]Segmentation: projection and character feature
Recognition: OCR
2012

Yahoo36%–89%[41]Segmentation: color filling
Combination: redundancy
Recognition: CNN
Postprocessing: DFS
2013

Microsoft5.56%[60]Different width/location segmenting and template matching2015
57.05%

reCAPT-CHA40.4%–94.3%[61] Segmentation: trichromatic code2015
Recognition: SVM

Yahoo57.3%–76.7%[7]Edge and fuzzy logic segmentation and recognition2015

Microsoft5%–77.2%[42]Segmentation: Log-Gabor filter
Combination: redundancy
Recognition: KNN
Postprocessing: DP search
2016

MSN27.1%–53.2%[48]Segmentation: different width
Recognition: BPNN
2016

Note. CNN: convolutional neural network, DP: dynamic programming, OCR: optical character recognition, DFS: depth first search, KNN: -nearest neighbor, BPNN: back-propagation neural network.