• 한국통신학회논문지
• /
• 제35권1C호
• /
• pp.24-32
• /
• 2010

본 논문에서는 버려지거나 없어진 객체를 검출하는 시스템에 대해 연구하였다. 전경과 배경을 분리한 다음, 정적인 영역에 대한 검출을 통하여 방치되거나 제거된 물체를 검출하였다. 정적인 영역에 대한 검출 방법을 제안하고 히스토그램의 비교를 통해 방치, 제거 정보를 추출하였다. 제안된 방법은 CCTV 카메라의 입력 영상에 대하여 PC 및 DSP 칩을 이용하여 실시간 처리를 하였으며 DSP칩을 활용하였기 때문에 수정이 용이하다. 제안된 시스템에 대한 성능을 검증하기 위해 저, 중, 고의 복잡도에 따라 실험하였으며, 신뢰성 있는 검증을 위해 각 10회의 반복 수행을 하였다. 실험 결과, 복잡도가 낮거나 보통인 경우는 높은 객체 변화 검출률을 보였으며 매우 혼잡한 경우에는 환경적인 요인의 이유로 검출률이 상대적으로 낮은 것을 확인할 수 있었다. 이 원인은 복잡도가 높아짐에 따라 검출률이 낮은 것은 이동하는 객체들로 인해 방치된 객체의 폐색이 반복되기 때문이었다. 향후 이러한 문제의 해결을 위해 매우 복잡한 환경에서의 폐색에 대한 추가적인 연구와 강건한 정적 영역의 판단 방법에 대해 연구할 것이다.

• 방송공학회논문지
• /
• 제16권3호
• /
• pp.462-470
• /
• 2011

본 논문에서는 실시간 영상감시 시스템에서 방치되거나 제거된 객체를 강인하게 검출하는 알고리듬을 제안한다. 방치되거나 없어진 객체는 사람이나 다른 움직임의 원인에 의해서 나타나기 때문에 추적을 기반으로 한 방법을 사용하였다. 가우시안 혼합 모델에 의해 전경과 배경을 분리하고 그림자 제거 알고리듬을 적용하였다. 모폴로지를 수행하여 주변 잡음을 제거하고 객체 구분을 명확히 하였다. 그리고 검출된 정지 객체를 방치되거나 사라지는 객체 중의 하나로 분류하였다. 추적방법을 적용함과 동시에 정지된 객체에 모니터링 시간까지 부여하여 검출된 객체가 다른 객체에 의해서 폐색되는 현상을 개선하였다. 영역성장기법을 이용하여 방치된 객체와 제거된 객체의 분류 성능을 높였다. 또한 제안된 시스템을 DSP 기반으로 설계하여 실시간 구현을 하였다. 실험을 통해 제안한 방법의 우수성을 입증하였다.

• 한국군사과학기술학회지
• /
• 제25권3호
• /
• pp.300-310
• /
• 2022

A vast amount of labeled data is required for deep neural network training. A typical strategy to improve the performance of a neural network given a training data set is to use data augmentation technique. The goal of this work is to offer a novel image augmentation method for improving object detection accuracy. An object in an image is removed, and a similar object from the training data set is placed in its area. An in-painting algorithm fills the space that is eliminated but not filled by a similar object. Our technique shows at most 2.32 percent improvements on mAP in our testing on a military vehicle dataset using the YOLOv4 object detector.

• 한국멀티미디어학회논문지
• /
• 제16권10호
• /
• pp.1171-1179
• /
• 2013

IAS는 대게 사용자가 자신의 취향에 맞는 음악을 직접 제작 및 편집 가능한 기능을 제공하는 서비스이다. SAOC는 낮은 전송률로 IAS가 가능한 다객체 오디오 코딩 기술이다. 하지만 SAOC 기법은 특정 객체를 제거하는 경우, 특히 보컬 객체를 제거하는 경우 배경음악에 보컬 객체의 하모닉이 남아있는 문제점이 있다. 그래서 본 논문은 하모닉 추출과 제거를 사용한 보컬 객체 제거 기법을 제안한다. 제안 하는 기법은 부호화기에서 추출한 하모닉 정보를 이용하여 복호화기에서 보컬 객체 신호를 다운믹스 신호에서 제거하는 기법이다. 하모닉 정보로써, 기본 주파수, MVF, 하모닉 크기를 사용한다. 성능평가로 객관적, 주관적 실험을 수행하였으며 모든 실험 결과를 통해 SAOC 기법보다 제안하는 기법이 우수함을 확인한다.

• Journal of Information Display
• /
• 제3권2호
• /
• pp.19-25
• /
• 2002

In this paper, a new target extraction algorithm is proposed, in which the coordinates of target are obtained adaptively by using the difference image information and the optical BPEJTC(binary phase extraction joint transform correlator) with which the target object can be segmented from the input image and background noises are removed in the stereo vision system. First, the proposed algorithm extracts the target object by removing the background noises through the difference image information of the sequential left images and then controlls the pan/tilt and convergence angle of the stereo camera by using the coordinates of the target position obtained from the optical BPEJTC between the extracted target image and the input image. From some experimental results, it is found that the proposed algorithm can extract the target object from the input image with background noises and then, effectively track the target object in real time. Finally, a possibility of implementation of the adaptive stereo object tracking system by using the proposed algorithm is also suggested.

• 대한임베디드공학회논문지
• /
• 제19권1호
• /
• pp.65-71
• /
• 2024

An indirect visual SLAM takes raw image data and exploits geometric information such as key-points and line edges. Due to various environmental changes, SLAM performance may decrease. The main problem is caused by dynamic objects especially in highly crowded environments. In this paper, we propose a robust feature-based visual SLAM, building on ORB-SLAM, via multi-channel dynamic objects estimation. An optical flow and deep learning-based object detection algorithm each estimate different types of dynamic object information. Proposed method incorporates two dynamic object information and creates multi-channel dynamic masks. In this method, information on actually moving dynamic objects and potential dynamic objects can be obtained. Finally, dynamic objects included in the masks are removed in feature extraction part. As a results, proposed method can obtain more precise camera poses. The superiority of our ORB-SLAM was verified to compared with conventional ORB-SLAM by the experiment using KITTI odometry dataset.

• 제어로봇시스템학회논문지
• /
• 제21권10호
• /
• pp.958-964
• /
• 2015

In this paper, an object recognition method based on the depth information from the RGB-D camera, Xtion, is proposed for an indoor mobile robot. First, the RANdom SAmple Consensus (RANSAC) algorithm is applied to the point cloud obtained from the RGB-D camera to detect and remove the floor points. Next, the removed point cloud is classified by the k-means clustering method as each object's point cloud, and the normal vector of each point is obtained by using the k-d tree search. The obtained normal vectors are classified by the trained multi-layer perceptron as 18 classes and used as features for object recognition. To distinguish an object from another object, the similarity between them is measured by using Levenshtein distance. To verify the effectiveness and feasibility of the proposed object recognition method, the experiments are carried out with several similar boxes.

• 전기학회논문지
• /
• 제67권4호
• /
• pp.578-585
• /
• 2018

In this paper, we propose a classification and theft detection system for human and livestock for various moving objects in a barn. To do this, first, we extract the moving objects using the GMM method. Second, the noise generated when extracting the moving object is removed, and the moving object is recognized through the labeling method. And we propose a method to classify human and livestock using model formation and color for the unique form of the detected moving object. In addition, we propose a method of tracking and overlapping the classified moving objects using Kalman filter. Through this overlap determination method, an event notifying a dangerous situation is generated and a theft determination system is constructed. Finally, we demonstrate the feasibility and applicability of the proposed system through several experiments.

• 한국실내디자인학회논문집
• /
• 제14권6호
• /
• pp.103-111
• /
• 2005

Since the Industrial Revolution in the 18th century, the production system adopted mechanization and mass production to popularize and standardize overall society. The architectural space removed basic desire of men on the decoration to neglect historical and regional continuity of the architecture and to make uniform designs, so that human sensibility and emotion were excluded. The architectural space had arbitrary and functionalism features in accordance with such a social change to display abstract space having no personality. The limitation expanded value of the space that was not lot possession and residence but for communication with men to express object factors. However, the object factors of the space were expressed in each factor consisting of either material factors or non-material factors. This study investigated interaction of the space by both expressive interaction and potential interaction to find out key words for making frames of the investigation and to examine cases comprehensively. The purpose of the study was to recover relations between the space and men in horizontal way and mutual communication and to present possibility integrating men and space.

• International Journal of Advanced Culture Technology
• /
• 제6권4호
• /
• pp.317-322
• /
• 2018

We use edge detection technique for the input image to extract the entire edges of the object in the image and then select only the edges that construct the outline of the object. By examining the positional relation between these pixels composing the outline, a simplified version of the outline of the object in the input image is generated by removing unnecessary pixels while maintaining the condition of connection of the outline. For each pixel constituting the outline, its direction is calculated by examining the positional relation with the next pixel. Then, we group the consecutive pixels with same direction into one and then change them to a line segment instead of a point. Among those line segments composing the outline of the object, a line segment whose length is smaller than a predefined minimum length of acceptable line segment is removed by merging it into one of the adjacent line segments. As a result, an outline composed of line segments of over a certain length is obtained through this process.