Color, Scale, and Rotation Independent Multiple License Plates Detection in Videos and Still Images

Table 2

Performance comparison of the proposed LP detection method with few of the competitive LP detection methods from the literature.

Algorithm and references	Data set	Number of images in data set	Types of vehicles present in data set	LP detection rate
CIP [3]	Proprietary	80	Cars	91.25%
ESM [4]	Proprietary	9825	Vans, trucks, cars	99.6%
FCC [5]	Proprietary	150	Not reported	95.3%
SCW [6]	Media-lab (proprietary)	1334	Vans, trucks, cars	96.5%
RTR [7]	Proprietary	400	Not reported	83.5%
RTR [7]	Caltech	1999	Cars	84.8%
SWHVP [8]	Proprietary	522	Motorcycles	97.55%
Two pass [9]	Proprietary	9026	Cars, trucks	97.16%
RELIP [10]	Proprietary	100	Not reported	97%
PVW [11]	Proprietary	410	Not reported	93.2%
VEDA [12]	Proprietary	664	Cars	91.65%
AOLP [13]	Media-lab	741	Vans, trucks, cars	92.1%
AOLP [13]	AOLP (proprietary)	2049	Cars and vans	93.33%
DIP-GA [14]	Proprietary	800	Vans, trucks, cars	98.75%
DIP-GA [14]	Media-lab	335	Vans, trucks, cars	97.61%
Proposed geometry-based clustering method	Media-lab	741	Vans, trucks, cars	97.3%
Proposed geometry-based clustering method	Proprietary	159	Cars, trucks, motorcycles	98.74%
Proposed geometry-based clustering method	Media-lab and proprietary	900	Vans, trucks, cars, motorcycles	97.56%
Proposed geometry-based clustering method	AOLP	2049	Cars and vans	93.7%