학술논문
차량 전면 영상을 이용한 차량의 색상과 조수석 탑승자 검출 / Vehicle Color and Passenger Detection Using Vehicle Front Images
Document Type
Dissertation/ Thesis
Author
Source
Subject
Language
Korean
Abstract
본 논문에서는 특정 차량 검색을 위해 차량 전면 영상을 이용하여 차량의 특징을 검출하는 방법을 제안한다. 제안하는 방법은 차량 전면 영상과 차량 번호판의 정보를 함께 이용하여 차량의 특징을 검출하는 것이다. 검출하고자 하는 차량의 특징은 색상 정보와 조수석 탑승자의 유무이다. 기존의 차량 색상 검출에 관한 연구는 차량 전체의 색상 정보나 차량의 넓은 영역에서의 색상 정보를 이용하는데, 이러한 방법은 차량의 색상이 포함되지 않은 유리창 같은 영역을 포함하기 때문에 오류가 일어난다. 따라서 본 논문에서는 번호판 위치 정보를 기준으로 추출한 관심영역(ROI)에서의 색상 정보와 색상의 히스토그램을 이용하는 방법에 대해 제안하여 기존 방법에서 발생하는 색상 인식 오류를 감소시켰다. 조수석 탑승자의 유무를 검출하는 방식도 마찬가지로 번호판 위치 정보를 기준으로 추출한 관심영역에서 에지를 검출하여 이를 분석하는 방법에 대해 제안한다.
In this paper, we propose a method for detecting the feature of a vehicle using a vehicle front image for a specific vehicle search. The proposed method is to detect the feature of the vehicle by using vehicle front image and the information of the vehicle license plate. A feature of the vehicle to be detected is the color information and the presence or absence of the passenger seat occupants. Existing research of color detection uses color information of the entire area of vehicle or a partial region with a large area of vehicle. The preferred method, an error occurs because the windows which do not include a region such as the color of the vehicle. However this paper, we propose a method for using the histogram of the color and the color of the region of interest (ROI) is extracted based on the license plate location. As a result reduced the color recognition errors in already known method. Also detection methods of presence or absence of the passenger seat occupants extracts a region of interest based on the license plate location. And we propose a method for analyzing the detected edge from the extracted region.
In this paper, we propose a method for detecting the feature of a vehicle using a vehicle front image for a specific vehicle search. The proposed method is to detect the feature of the vehicle by using vehicle front image and the information of the vehicle license plate. A feature of the vehicle to be detected is the color information and the presence or absence of the passenger seat occupants. Existing research of color detection uses color information of the entire area of vehicle or a partial region with a large area of vehicle. The preferred method, an error occurs because the windows which do not include a region such as the color of the vehicle. However this paper, we propose a method for using the histogram of the color and the color of the region of interest (ROI) is extracted based on the license plate location. As a result reduced the color recognition errors in already known method. Also detection methods of presence or absence of the passenger seat occupants extracts a region of interest based on the license plate location. And we propose a method for analyzing the detected edge from the extracted region.