A Survey on Breaking Technique of Text-Based CAPTCHA

Table 2

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

Example	Source	Success rate	Reference	Breaking method	Year
	Gimpy-r	78%	[57]	Segmentation: character gap	2004
				Recognition: distortion evaluation
	EZ-Gimpy	97.9%	[58]	Segmentation: connected region	2004
				Recognition: distortion evaluation
	Captcha-service	100%	[34]	Segmentation: vertical projection; Recognition: statistical character pixels	2007
	Ego-share	92.2%	[5]	Segmentation: connected region	2009
				Recognition: SVM
	Ge-Captcha	100%	[59]	CW-SSIM	2010

Note. SVM: support vector machine, CW-SSIM: complex wavelet based structural similarity.