• 한국멀티미디어학회논문지
• /
• 제24권2호
• /
• pp.305-311
• /
• 2021

One of the highly reliable object tracking methods is to trace objects by associating objects detected by deep learning. The detected object is represented by a rectangular box. The box has information such as location and size. Since the tracker has motion information of the object in addition to the location and size, knowing additional information about the motion of the detected box can increase the reliability of object tracking. In this paper, we present a new method of reliably estimating the moving direction of the detected object in underground parking lot. First, the frame difference image is binarized for detecting motion energy, change due to the object motion. Then, a cumulative binary image is generated that shows how the motion energy changes over time. Next, the moving direction of the detected box is estimated from the accumulated image. We use a new cost function to accurately estimate the direction of movement of the detected box. The proposed method proves its performance through comparative experiments of the existing methods.

• 한국정보처리학회:학술대회논문집
• /
• 한국정보처리학회 2016년도 추계학술발표대회
• /
• pp.506-509
• /
• 2016

We present real-time pedestrian detection that exploit accuracy of Faster R-CNN network. Faster R-CNN has shown to success at PASCAL VOC multi-object detection tasks, and their ability to operate on raw pixel input without the need to design special features is very engaging. Therefore, in this work we apply and adjust Faster R-CNN to single object detection, which is pedestrian detection. The drawback of Faster R-CNN is its failure when object size is small. Previously, small sized object problem was solved by Scale-aware Network. We incorporate Scale-aware Network to Faster R-CNN. This made our method Scale-aware Faster R-CNN (DF R-CNN) that is both fast and very accurate. We separated Faster R-CNN networks into two sub-network, that is one for large-size objects and another one for small-size objects. The resulting approach achieves a 28.3% average miss rate on the Caltech Pedestrian detection benchmark, which is competitive with the other best reported results.

• 한국의류학회지
• /
• 제28권5호
• /
• pp.626-636
• /
• 2004

The purpose of this study was to show a theoretical system of expressional area, the characteristics and the effects which is applied to fashion illustration by metaphor theory. The theoretical system of expressional area was analyzed by category analysis and 150 fashion illustrations from 1900 to 1999 were analyzed by contents analysis. The results of this study were as follows: Metaphor in fashion illustration was categorized to replacement, parody, heterogeneous combination, surrealism, magnification and reduction of size, optical illusion, juxtaposition and distortion of form. In detail, First, replacement was showed omission of form, non-object color, texture. Second, parody was showed using the part in artistic work, differentiation of original work. Third, heterogeneous combination was showed unreal body combining between animal and plant. Fourth, surrealism was showed creation of object which is impossible to present. Fifth, magnification and reduction of size was showed bizarre magnification of part of body, size of clothing. Sixth, optical illusion was showed ambiguity of object because of creation of new form. Seventh, juxtaposition was showed the parallel of contradicting idea and change of meaning between heterogeneous objects. Eighth, the distortion of form was showed grotesque distortion of part of body and disgusting object.

• 한국전자파학회논문지
• /
• 제8권5호
• /
• pp.534-544
• /
• 1997

Levenberg-Marquardt (LMA)와 유전 알고리즘(GA)을 결합한 새로운 잡종알고리틈을 반복적으로 사용하여, 비용함수의 실 극소값(global minimum)을 주는 2차원 강산란체의 유전율 분포를 재구성한다. 비용함수에 사용되는 산란파는 원통형 각모드로 전개되며, 이 중 유효 전파모드만이 이용된다. 유효 전파모드만을 사용하여 비용함수를 정의함으로써 주어진 산란체를 재구성하는데 필요한 입사파 개수의 최소값이 공식화된다. 수치해석 결과로부터,LMA는 수렴 속도가 빠르나 강산란체를 재구성할 수 없고, GA는 강산란체의 재구성은 가능하나 수렴 속도가 느린 반면, 결합 알고리즘을 이용하는 역산란 방법은 LMA와 GA의 장점만을 취합한 방법임이 입증된다.

• 한국기계가공학회지
• /
• 제16권5호
• /
• pp.134-140
• /
• 2017

Shape-from-shading (SFS) or stereo vision algorithms can be utilized to measure the shape of an object with imaging techniques for effective sensing in non-contact measurements. SFS algorithms could reconstruct the 3D information from a 2D image data, offering relatively comprehensive information. Meanwhile, a stereo vision algorithm needs several feature points or lines to extract 3D information from two 2D images. However, to measure the size of an object with a freeform surface, the two algorithms need some additional information, such as boundary conditions and grids, respectively. In this study, a stereo vision scheme using the depth information obtained by shape-from-shading as patterns was proposed to measure the size of an object with a freeform surface. The feasibility of the scheme was proved with an experiment where the images of an object were acquired by a CCD camera at two positions, then processed by SFS, and finally by stereo matching. The experimental results revealed that the proposed scheme could recognize the size and shape of freeform surface fairly well.

• 정보처리학회논문지B
• /
• 제8B권5호
• /
• pp.539-548
• /
• 2001

본 연구에서는 배경과 구분되는 이동물체를 추적하기 위한 방법으로 부분 외곽선 정보를 이용한 이동물체 추적 알고리즘을 제안하였다. 이동물체의 추적은 이동물체의 외곽선을 검출한 다음 외곽선 정보를 이동물체의 특징으로 정하여 추적하는 알고리즘을 사용하였다. 먼저 이동물체 외곽선 정보를 이용하여 연속한 동영상 입력에 대하여 속 BMA(Block Matching Algorithm)을 이용하여 움직임 벡터를 추출하고 움직임 벡테를 기초로 이동물체를 추출한다. 다음은 이동물체 초기 특징 벡테 생성단계로서 이동물체에 대한 외곽선을 추출한다. 이동물체의 외곽선 영역 중 상하좌우의 외곽선 일부분을 특징벡터로 정한다. 다음은 추적단계로 이전 프레임에서 얻은 특징벡터를 이용하여 현재 프레임에서 이동물체의 추적을 수행하였다. 제안된 알고리즘에 대하여 실제영상을 가지고 이동물체추적 모의 실험을 수행한 결과 기존 능동 윤곽선 추적알고리즘은 물체 외곽선 전체를 추적하기 때문에 물체의 외곽선 길이에 따라 처리시간이 변화하지만 제안된 알고리즘은 이동물체의 외곽선 영역을 특징정보로 하여 추적하기 때문에 추적연산이 간단하였다. 제안된 이동물체 추적알고리즘 중 이동벡터를 추출하는 BMA 연산은 기존 알고리즘 보다 연산량이 약 39%감소였으며, 상하 좌우 외곽선 정보를 이용하여 이동물체를 추적한 결과 추적오차는 특징벡터의 크기가 [$10{\times}5$]일 때 검색오차가 2화소 이하로 양호하게 나타났다. 또한 기본 능동 윤ㅅ곽선 축적알고리즘은 물체 외곽선 크기에 따른 처리시간이 변화하지만 제안된 알고리즘은 특징벡터의 크기가 일정하기 때문에 동일한 처리시간이 필요하였다.

• 한국조명전기설비학회:학술대회논문집
• /
• 한국조명전기설비학회 2004년도 춘계학술대회 논문집
• /
• pp.479-485
• /
• 2004

The purpose of this study is to develop a practical image inspection system that could recognize it correctly, endowing flexibility to the productive field, although the same object for work will be changed in the size and rotated. In this experiment, it selected a fighter, rotating the direction from $30^{\circ}\;to\;45^{\circ}$ simultaneously while changing the size from 1/4 to 1/16, as an object inspection without using another hardware for exclusive image processing. The invariant moments, Hu has suggested, was used as feature vector moment descriptor. As a result of the experiment the image inspection system developed from this research was operated in real-time regardless of the chance of size and rotation for the object inspection, and it maintained the correspondent rates steadily above from 94% to 96%. Accordingly, it is considered as the flexibility can be considerably endowed to the factory automation when the image inspection system developed from this research is applied to the productive field.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2000년도 하계종합학술대회 논문집(4)
• /
• pp.202-205
• /
• 2000

In this paper we propose an algorithm that detects, tracks a moving object, and classify whether it is human from the video clip captured under the fixed video camera. It detects the outline of the moving object by finding out the local maximum points of the modulus image, which is the magnitude of the motion vectors. It also estimates the size and the center of the moving object. When the object is detected, the algorithm discriminates whether it is human by segmenting the face. It is segmented by searching the elliptic shape using Hough transform and grouping the skin color region within the elliptic shape.

• 대한전자공학회논문지
• /
• 제27권6호
• /
• pp.861-870
• /
• 1990

A method of recognizing objects is proposed that uses a concept of modified incremental circle transform. The modified incremental circle transform, which maps bundaries of an object into an unit circle, represnets efficiently the shape of the boundaries detected in digitized binary images of the objects. It is proved that modified incremental circle transform of object, which is invariant under object translation, rotation, and size, can be used as feature information for recognizing two dimensional polygonal object efficiently.

• 한국정보과학회논문지:정보통신
• /
• 제31권3호
• /
• pp.305-313
• /
• 2004

웹 객체의 저장과 처리를 위한 캐시의 사용이 증대하고 있으며, 캐시 저장영역의 효율적인 관리를 위한 많은 연구가 활발히 이루어지고 있다. 웹 캐싱 기법은 전통적인 기법과 차이가 있다. 특히 웹 캐싱의 처리 단위인 웹 객체의 이질성과, 시간에 따른 웹 객체 참조특성 변화는 기존 기법들의 성능을 감소시키는 중대한 원인이 되고 있다. 본 연구에서는 새로운 웹 캐싱 기법인 ACASH(the Adaptive Caching Algorithm with Size Heterogeneity)를 제안하였다. ACASH는 웹 객체와 캐시 영역을 이질성을 기반으로 분할 관리함으로서 객체의 이질성 편차를 줄였고, 시간의 흐름에 따른 객체 참조 특성의 변화를 적응적으로 반영하고 있다. 또한 객체의 이질성을 고려한 두 개의 실험 모델에 대해, 기존의 대체 기법들과 비교 실험을 통해 ACASH의 우수성을 확인하였다.