Analisis Deteksi Plat Nomor Kendaraan Menggunakan Metode Connected ComponentLabelling dan Template Matching
Penulis/Author
Prof. Ir. Hanung Adi Nugroho, S.T., M.Eng., Ph.D., IPM., SMIEEE. (1); Dr.Eng. Ir. Sunu Wibirama, S.T., M.Eng., IPM. (2)
Tanggal/Date
2015
Abstrak/Abstract
As well as the name of humans, license plate can
also be seen as the identity of a certain vehicle. The plate
recognition system is needed to noting and tracking a vehicle in a
certain area. In this experiment, the main objective is improving
the license plate recognition algorithm by considering the
prespective rectification and analyzing some of the methods and
parameters in license plate detection.
In this experiment, license plate detection uses the
connected component labelling method to detects the plate and
segments the character then using template matching method to
recognizes the character. Some operators of the edge detection,
binarization methods, and Gaussian filter variances will be
compared in this experiment.
The result shows that the designed system can detect
and recognize the vehicle license plate by 91.4% for 1 meter
distance and 83% for 3 meters distance. Planar Homography is
used to reduce the perspective distortion and increases the
accuracy by 3.5%
