Journal of the Korean Institute of Telematics and Electronics S (전자공학회논문지S)
- Volume 36S Issue 6
- /
- Pages.76-86
- /
- 1999
- /
- 1226-5837(pISSN)
Extraction of Car Number Plate Using Wavelet Transform
Wavelet 변환을 이용한 차량 번호판 영역 추출
- Hwang, Woon-Joo (Dept. of Electronic Engineering, University of Inchon) ;
- Park, Sung-Wook (Dept. of Electronic Engineering, University of Inchon) ;
- Park, Jong-Wook (Dept. of Electronic Engineering, University of Inchon)
- Published : 1999.06.01
Abstract
In this paper, it is shown that the car number plate are segmented and extracted more efficiently by using wavelet transform. A car image is decomposed by wavelet transform, and the high frequency image of the decomposed image are selected as feature images. Three selected feature images are synthesized of a single feature image, and a region including the plate is segmented by the correlation coefficient between the feature image and the synthesized image. For segmented plate region, the car plate region is extracted by deciding the Y-axis region composed by vertical region, the car plate region is extracted by deciding the Y-axis region composed by vertical histogram and the X-axis region composed by the variance histogram. Some experiment results of the various image and shown. It has been shown from the results with the high rate of 96% that the car number plates can be segmented and extracted more extractly and efficiently than converntional method.
본 논문은 웨이브렛 변환을 이용하여, 차량 번호판 영역을 추출하는 새로눙 방법을 제안한다. 차량 영상을 웨이브렛 변환에 의하여 분해하고 분해된 각 영상으로부터 생성되는 고주파 영상들은 특징 영상으로 선정한다. 선정된 3개의 특징 영상들을 하나의 특징 영상으로 합성하고, 번호판 위치 영역을 검출한다. 검출된 영역에 대하여 수직 방향 히스토그램에 의해 Y축 영역을 결정하고, 수평 방향 히스토그램의 분산 히스토그램을 구성하여 X축 영역을 결정하므로써의 번호판 영역을 추출한다. 또한 제안한 방법의 타당성을 보이기 위하여, 다양한 국내
Keywords