• 한국컴퓨터정보학회논문지
• /
• 제26권9호
• /
• pp.49-55
• /
• 2021

본 논문에서는 실제 도로에서 기울어진 촬영 각도로 인하여 회전된 차량 번호판을 정확하게 탐지하기 위하여 객체 세그먼테이션(object segmentation)을 이용하는 개선된 2-단계 차량 번호판 탐지 모델을 제안한다. 기존 연구에서 제안한 3-단계 차량 번호판 탐지 파이프라인 모델은 차량 번호판이 많이 기울어져 있을수록 탐지 정확도가 낮아지는 문제가 있다. 이를 해결하기 위해서 기존의 3-단계 모델에서 사각형 형태만으로 차량 후보 영역과 차량 번호판 후보 영역을 인식하는 전위 2개의 처리 단계 대신에 임의의 형태로 객체 탐지가 가능한 객체 세그먼테이션을 이용하는 하나의 단계로 대체함으로써 탐지 과정을 단순화하였으며 궁극적으로는 임의의 형태로 기울어진 차량 이미지에 대해서도 탐지 성능을 개선하였다. 기울어진 차량 번호판 이미지를 대상으로 실시한 차량 번호판 탐지 모델의 정확도 분석 실험 결과에 의하면 기존의 3-단계 차량 번호판 탐지 모델보다 제안된 2-단계 기법이 탐지 과정을 단순화하였음에도 최대 약 20%의 탐지 정확도를 개선할 수 있는 것으로 분석되었다.

• 스마트미디어저널
• /
• 제9권4호
• /
• pp.134-143
• /
• 2020

자동차 번호판 인식 기술은 도로의 교통상황 통제, 과속차량 단속, 도주 차량의 추적 등 현대 교통 시설 및 교통 안전망을 책임지고 있는 핵심 기술 중 하나이다. 이 기법은 과거에도 연구되었던 분야였으나 최근 딥러닝 기술의 발전으로 다양한 기법들을 적용하여 향상된 성능을 보이는 분야이며, 크게 자동차 번호판 검출과 번호판 인식으로 나뉜다. 본 연구에서는 다양한 객체 검출 모델과 WPOD-Net(Warped Planar Object Detection Network) 모델을 활용하여 자동차 번호판 검출 성능을 향상시키기 위한 실험을 진행하였으며, 객체 검출 모델을 활용하여 번호판을 검출하는 기존 방식들 대신 차량을 검출한 다음 번호판을 검출하는 방식을 택하여 정확도를 높였다. 특히 Super-Resolution 기법 중 하나인 Fast-SRGAN 모델을 활용하여 이미지 내에 존재하는 노이즈를 제거하는 처리를 통해 최종 성능을 향상시켰다. 결과적으로 92.38%에서 96.72%로 선행 연구 대비 평균 4.34% 향상된 성능이 실험을 통해 확인되었다.

• 한국IT서비스학회지
• /
• 제20권6호
• /
• pp.97-105
• /
• 2021

Recently, with the development of IT technology, unmanned systems are being introduced in many industrial fields, and one of the most important factors for introducing unmanned systems in the automobile field is vehicle licence plate recognition(VLPR). The existing VLPR algorithms are configured to use image processing for a specific type of license plate to divide individual areas of a character within the plate to recognize each character. However, as the number of Korean vehicle license plates increases, the law is amended, there are old-fashioned license plates, new license plates, and different types of plates are used for each type of vehicle. Therefore, it is necessary to update the VLPR system every time, which incurs costs. In this paper, we use an object detection algorithm to detect character regardless of the format of the vehicle license plate, and apply a handwritten Hangul recognition(HHR) algorithm to enhance the recognition accuracy of a single Hangul character, which is called a Hangul unit. Since Hangul unit is recognized by combining initial consonant, medial vowel and final consonant, so it is possible to use other Hangul units in addition to the 40 Hangul units used for the Korean vehicle license plate.

• 한국정밀공학회:학술대회논문집
• /
• 한국정밀공학회 2002년도 춘계학술대회 논문집
• /
• pp.229-232
• /
• 2002

This paper describes the development of robot's finger 3-axis force sensor that detects the Fx, Fy, and Fz simultaneously fur stably grasping an unknown object. In order to safely grasp an unknown object using the robot's fingers, they should detect the force of gripping direction and the force of gravity direction, and perform the force control using the detected farces. The 3-axis force sensor that detects the Fx, Fy, and Fz simultaneously should be used for accurately detecting the weight of an unknown object of gravity direction. Thus, in this paper, robot's finger for stably grasping an unknown object is developed. And, the 3-axis farce sensor that detects the Fx, Fy, and Fz simultaneously fur constructing a robot's finger is newly modeled using several parallel-plate beams, and is fabricated. Also, it is calibrated, and evaluated.

• International Journal of Precision Engineering and Manufacturing
• /
• 제4권3호
• /
• pp.12-19
• /
• 2003

This paper describes the design of a robot's hand with two fingers for stably grasping an unknown object, and the development of a 3-axis force sensor for which is necessary to constructing the robot's fingers. In order to safely grasp an unknown object using the robot's fingers, they should measure the forces in the gripping and in the gravity directions, and control the measured forces. The 3-axis force sensor should be used for accurately measuring the weight of an unknown object in the gravity direction. Thus, in this paper, the robot's hand with two fingers for stably grasping an unknown object is designed, and the 3-axis force sensor is newly modeled and fabricated using several parallel-plate beams.

• 한국콘크리트학회:학술대회논문집
• /
• 한국콘크리트학회 2006년도 춘계학술발표회 논문집(I)
• /
• pp.486-489
• /
• 2006

This paper presents a newly proposed object-oriented finite element framework and its applications on dynamic analysis of plate girder bridge. The developed framework supports various types of finite elements, materials, constraints, loads, and solution methods. One major feature different from other object-oriented finite element programs is that static model and dynamic state can be easily read from or written to a file. In addition, the framework supports efficient DOF pattern handling for a node connecting elements with different DOF patterns, new multi-point constraint handling, and various scripting languages for easy use of the library. In order to show the applicability to dynamic analysis, dynamic moving load analysis on plate girder bridge is performed.

• 대한기계학회:학술대회논문집
• /
• 대한기계학회 2003년도 춘계학술대회
• /
• pp.1548-1553
• /
• 2003

An experimental study of the absorption process of water vapor into a lithium bromide solution was performed. For the purpose of development of high performance absorption chiller/hater utilizing lithium bromide solution as working fluid, it is the most effective to improve the performance of absorber with the largest heat transfer area of the four heat exchangers. The experimental apparatus was composed of a plate type absorber which can increase the heat exchange area per unit volume to investigate more detail characteristics instead of the conventional type, horizontal tube bundle type. The size of plate absorbers were made for $0.4m{\times}0.6m$ and the design object of a refrigeration capacity was lRT. In this experiment, three kind plate absorbers which were flat plate, dimple plate and groove plate were used. The results were less than the design object values, that is, the refrigeration capacity was about $0.3{\sim}0.4RT$ and the overall heat transfer coefficient was $500{\sim}600kcal/m^2h^{\circ}C$ at the standard conditions.

• 한국산업융합학회 논문집
• /
• 제24권5호
• /
• pp.555-562
• /
• 2021

In this paper, we propose a sorting method for extracting accurate license plate information, which is currently used in Korea, after detecting objects using YOLO. We propose sorting methods for the five types of vehicle license plates managed by the Ministry of Land, Infrastructure and Transport by classifying the plates with the number of lines, Korean characters, and numbers. The results of experiments with 5 license plates show that the proposed algorithm identifies all license plate types and information by focusing on the object with high reliability score in the result label file presented by YOLO and deleting unnecessary object information. The proposed method will be applicable to all systems that recognize license plates.

• 한국CDE학회논문집
• /
• 제14권2호
• /
• pp.129-136
• /
• 2009

Generally, curved surfaces of ship hull are deformed by flame bending (line heating), multi-press forming, and die-less forming method. The forming methods generate the required in-plane/bending strain or displacement on the flat plate to make the curved surface. Multi-press forming imposes the forced displacements on the flat plate by controlling the position of each pressing points based upon the shape difference between the unfolded flat plate and the curved object shape. The flat plate has been obtained from the unfolding system that is independent of the ship CAD. Apparently, the curved surface and the unfolded-flat surface are expressed by different coordinate systems. Therefore, one of the issues is to find a registration of the unfolded surface and the curved shape for the purpose of minimum amount of forming works by comparing the two surfaces. This paper presents an efficient algorithm to get an optimized registration of two different surfaces in the multi-press forming of ship hull plate forming. The algorithm is based upon the ICP (Iterative Closest Point) algorithm. The algorithm consists of two iterative procedures including a transformation matrix and the closest points to minimize the distance between the unfolded surface and curved surfaces. Thereby the algorithm allows the minimized forming works in ship-hull forming.

• 디지털산업정보학회논문지
• /
• 제18권4호
• /
• pp.1-11
• /
• 2022

Recently, research on license plate recognition, which is a core technology of an intelligent transportation system(ITS), is being actively conducted. In this paper, we propose a method to extract numbers and characters from low-quality license plate images by applying the YOLOv4 algorithm. YOLOv4 is a one-stage object detection method using convolution neural network including BACKBONE, NECK, and HEAD parts. It is a method of detecting objects in real time rather than the previous two-stage object detection method such as the faster R-CNN. In this paper, we studied a method to directly extract number and character regions from low-quality license plate images without additional edge detection and image segmentation processes. In order to evaluate the performance of the proposed method we experimented with 500 license plate images. In this experiment, 350 images were used for training and the remaining 150 images were used for the testing process. Computer simulations show that the mean average precision of detecting number and character regions on vehicle license plates was about 93.8%.