I.N. Trapeznikov – Postgraduate student, Yaroslavl State University. E-mail: email@example.com
A.L. Priorov – Dr.Sc. (Eng), Associate Professor, Yaroslavl State University. E-mail: firstname.lastname@example.org
A.A. Noskov – Postgraduate student, Yaroslavl State University. E-mail: email@example.com
E.A. Aminova – Postgraduate student, Yaroslavl State University. E-mail: firstname.lastname@example.org
Actually, one of the development directions of television signals analysis systems in the various applications is to identify and recognize objects in a digital image. Considered algorithms are quite appropriate in the automatic automobile license numbers recognition.
The goal of this article is detection of license numbers on the digital image issues as the main part of the recognition systems. The methods of the objects detection on proposed shapes such as the automobile license numbers are presented. The combining the considered approaches of computing landmarks of image and machine learning methods prevents the usage a priori information about the properties of license plate such as a size, an aspect ratio, etc. The developed algorithm uses a multi-pronged approach to determine the landmarks and the computation of descriptors areas of interest. Simultaneously, there are no strict limitations on the size of the automobile license plates, turn angle, etc.
Each person determines the license plates on the "normal" language; therefore, it is necessary to find out the definition of the automobile license plate based on descriptors that would be understandable for the system. The distinct landmark quantifies the angle based on the analysis of the Harris eigenvalues. After binarization algorithm Harris response card an image which is composed of a few connected regions is formed. In order to determine which area is the automobile license plate within the algorithm solves the issue of mentioned above regions classification by using the detecting anomalies algorithm. Characteristic features are calculated correspond to the histogram of oriented gradients (HOG) for this. The predictive ability of the constructed model is quite clearly characterized by the ROC-curves.
This article contains an analysis of the scientific and methodological object detection methods and machine learning algorithms including the capability of practical applying for the automatic automobile license plate recognition.
Dam'yanovski V. Bibliya videonablyudeniya. Cifrovy'e
i setevy'e texnologii. M.: Aj-E's-E's press. 2006.
Martinsky O. Algorithmic and mathematical principles
of automatic number plate recognition systems // B. Sc. Thesis, Brno
University of Technology. 2007.
Priorov A.L., Voloxov V.A., Sergeev E.V., Mochalov I.S.
Parallel'naya proczedura fil'traczii izobrazhenij na osnove analiza glavny'x
komponent nelokal'noj obrabotki // E'lektromagnitny'e volny' i e'lektronny'e
sistemy'. 2012. № 11. S. 64‑70.
Priorov A., Tumanov K., Volokhov V., Sergeev E., Mochalov
I. Applications of Image Filtration Based on Principal Component Analysis and
Nonlocal Image Processing // IAENG International Journal of Computer Science.
40:2. R. 62‑80.
Priorov A., Volokhov V., Sergeev E.,
Mochalov I., Tumanov K. Parallel Filtration Based on Principle Component
Analysis and Nonlocal Image Processing // Proc. International MultiConference
of Engineers and Computer Scientists 2013. Hong Kong, 2013. V. 1. P. 430-435.
Harris C., Stephens M. A combined corner and edge
detection // Plessley research Roke Manor. UK. 1988.