• Journal of Multimedia Information System
• /
• 제9권3호
• /
• pp.183-190
• /
• 2022

Image-based 3D object detection is one of the important and difficult problems in autonomous driving and robotics, and aims to find and represent the location, dimension and orientation of the object of interest. It generates three dimensional (3D) bounding boxes with only 2D images obtained from cameras, so there is no need for devices that provide accurate depth information such as LiDAR or Radar. Image-based methods can be divided into three main categories: monocular, stereo, and multi-view 3D object detection. In this paper, we investigate the recent state-of-the-art models of the above three categories. In the multi-view 3D object detection, which appeared together with the release of the new benchmark datasets, NuScenes and Waymo, we discuss the differences from the existing monocular and stereo methods. Also, we analyze their performance and discuss the advantages and disadvantages of them. Finally, we conclude the remaining challenges and a future direction in this field.

• 한국정보처리학회논문지
• /
• 제7권3호
• /
• pp.992-999
• /
• 2000

Textile dimension inspection is one of the basic issues in the textile dyeing and finishing industry. And also, it a plays an important role in the quality control of total fabric products. In this paper, we implement a real-time textile dimension inspection system which detects various real defects, defects positions of textile and the density of textiles. The proposed method consists of textile density measurement algorithms with zone-occurrence features from subband image which detect various types of real defects. The performance of the proposed method is tested with a number of real textile samples with 10 types of defects and three basic structures of textile. By the dimension inspection of textile at continuous stages in the fabrication process, it is possible to measure the density of textile up to 150m/min and to detect the defect of textile at real time within $\pm$1% error percentages. And also it can be monitored the condition of textile throughout at all the significant working process and can be improved textile quality.

• 한국생산제조학회지
• /
• 제8권3호
• /
• pp.92-99
• /
• 1999

The inspection unit which is developed and used in this study, is processed the shape data from the CCD camera to seek welding bite section shape, and then calculated as a real dimension from measuring the value of each inspection item. The reason of measuring with the real in this study is came out from the image method which used for a long time, which is extricated the characteristic as the dimension of pixel by recognize pixel. The measurement method of the section shape is that we decide the thresholding value after we drew the histogram to binarizate the object. After that, we make flat the object to get rid of the noise and measure the shape of welded part through the boundarization of the object. The shape measurement is that measure the value of the welding part to adapt the actual operation program from using the ratio between the actual dimension of the standard specimen and the dimension of image, to measure the ratio between the actual product and the camera image. The inspection algorithm which estimates the quality of welded product is developed and also, the software GUI(Graphic User Interface) which processes the automatic test function of the inspection system is developed. We make the foundation of the inspection automatic system and we will help to apply other welding machine.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 2000년도 제15차 학술회의논문집
• /
• pp.149-149
• /
• 2000

We experimented on AGV driving test with color CCD camera which is setup on it. This paper can be divided into two parts. One is image processing part to measure the condition of the guideline and AGV. The other is part that obtains the reference steering angle through using the image processing parts. First, 2 dimension image information derived from vision sensor is interpreted to the 3 dimension information by the angle and position of the CCD camera. Through these processes, AGV knows the driving conditions of AGV. After then using of those information, AGV calculates the reference steering angle changed by the speed of AGV. In the case of low speed, it focuses on the left/right error values of the guide line. As increasing of the speed of AGV, it focuses on the slop of guide line. Lastly, we are to model the above descriptions as the type of PID controller and regulate the coefficient value of it the speed of AGV.

• 한국지리정보학회지
• /
• 제7권4호
• /
• pp.24-33
• /
• 2004

원격탐사(remote sensing) 기술의 비약적인 발전과 함께 위성 영상데이터의 분광대역수가 급속히 증가하고 있다. 대역수의 증가로 영상데이터량이 급격히 증가하게 되고, 이에 따라 이들 데이터를 처리하기 위해서는 처리속도가 빠른 영상처리 기술이 필요하게 되었다. 분광 대역 수를 줄여 빠르게 처리하는 한가지 방법으로 널리 사용되고 있는 것이 주성분 변환법이다. 본 논문에서는 주성분 변환법에 대한 처리과정에 대해 논하였으며, 위성 영상데이터를 주성분 변환한 결과인 주성분 영상데이터를 분석하였다. 분석결과 실험 영상데이터의 경우, 3개의 주성분($PC_1$, $PC_2$, $PC_3$)의 누적 백분율 분산 값이 99.1%로 이는 3개의 주성분이 거의 대부분의 정보를 가지고 있음을 알 수 있었다. 3개의 주성분 영상데이터만을 사용한다면 데이터 저장을 위한 메모리 용량이나 데이터 전송시간 및 처리시간을 크게 감소시킬 수 있다. 또한 본 논문에서는 주성분 영상데이터를 최대유사분류법과 신경회로망을 이용한 다층 퍼셉트론 분류법으로 분류하고 결과를 평가한 후, 주성분 변환법이 갖는 차원축소 효과를 분석하였다. 분석결과 주성분 3개를 사용한 분류결과와 주성분 6개를 사용한 분류결과간의 분류정답률이 크게 차이가 나지 않았다. 이는 분류에 사용하는 영상데이터 수를 6개 차원에서 3개 차원으로 줄여도 비슷한 분류성능을 얻을 수 있음을 확인할 수 있었다.

• International Journal of Advanced Culture Technology
• /
• 제6권3호
• /
• pp.142-150
• /
• 2018

This paper describes the algorithm that lowers the dimension, maintains the object recognition and significantly reduces the eigenspace configuration time by combining the edge orientation histogram and principle component analysis. By using the detected object region as a recognition input image, in this paper the object recognition method combined with principle component analysis and the multi-layer network which is one of the intelligent classification was suggested and its performance was evaluated. As a pre-processing algorithm of input object image, this method computes the eigenspace through principle component analysis and expresses the training images with it as a fundamental vector. Each image takes the set of weights for the fundamental vector as a feature vector and it reduces the dimension of image at the same time, and then the object recognition is performed by inputting the multi-layer neural network.

• 한국항해항만학회:학술대회논문집
• /
• 한국항해항만학회 2000년도 추계학술대회논문집
• /
• pp.207-217
• /
• 2000

In this paper, we describe an image processing algorithm which is able to recognize the road lane. This algorithm performs to recognize the interrelation between AGV and the other vehicle. We experimented on AGV driving test with color CCD camera which is setup on the top of vehicle and acquires the digital signal. This paper is composed of two parts. One is image preprocessing part to measure the condition of the lane and vehicle. This finds the information of lines using RGB ratio cutting algorithm, the edge detection and Hough transform. The other obtains the situation of other vehicles using the image processing and viewport. At first, 2 dimension image information derived from vision sensor is interpreted to the 3 dimension information by the angle and position of the CCD camera. Through these processes, if vehicle knows the driving conditions which are angle, distance error and real position of other vehicles, we should calculate the reference steering angle.

• 한국콘크리트학회:학술대회논문집
• /
• 한국콘크리트학회 2001년도 봄 학술발표회 논문집
• /
• pp.333-338
• /
• 2001

Fractal geometry is a non-Euclidean geometry which has been developed to quantitative analysis irregular or fractional shapes. Fractal dimension of irregular surface has fractal values ranging from 2 to 3 and of irregular line profile has fractal values ranging from 1 to 2. In this paper, quantitative analysis of crack growth patterns during the fracture processing of fiber-reinforced cement composites based on fractal geometry. The fracture behaviors of fiber reinforced mortar beams subjected to three-point loading in flexure. The beams all had a single notch depth, but varing volume fractions of polypropylene, cellulose fibers. The crack growth behaviors, as observed through the image processing system, and the box counting method was used to determine the fractal dimension, Df. The results showed that the linear correlation exists between fractal dimension and fracture energy of the fiber reinforced cement mortar.

• 제어로봇시스템학회:학술대회논문집
• /
• 제어로봇시스템학회 1989년도 한국자동제어학술회의논문집; Seoul, Korea; 27-28 Oct. 1989
• /
• pp.700-704
• /
• 1989

A robotic system with a 3-dimensional profile measuring sensor is developed in order to measure the complicated shape of the target body. Due to this 3-dimensional profile measuring sensor, a computer is able to adjust the posture of the robot hand so that complicated global profile of the target body can be recognized after several measurements from the variant directions. In order to enable fast data processing, a digital signal processor and a look-up table is introduced.

• 대한전기학회:학술대회논문집
• /
• 대한전기학회 2001년도 하계학술대회 논문집 D
• /
• pp.2202-2205
• /
• 2001

This paper is composed of two parts. One is image preprocessing part to measure the condition of the lane and vehicle. This finds the information of lines using RGB ratio cutting algorithm, the edge detection and Hough transform. The other obtains the situation of other vehicles using the image processing and viewport. At first, 2 dimension image information derived from vision sensor is interpreted to the 3 dimension information by the angle and position of the CCD camera. Through these processes, if vehicle knows the driving conditions which are lane angle, distance error and real position of other vehicles, we should calculate the reference steering angle by steering controller.