• 한국HCI학회:학술대회논문집
• /
• 2007.02a
• /
• pp.725-730
• /
• 2007

본 논문에서는 인체 추적에 필요한 인체의 각 부위들을 구성요소로 각각 검출하여 연결하는 인체 모델을 통해 각 구성요소를 개별적으로 추정하게 된다. 여기서 인체의 구성요소 중 동작 추적에 가장 필요한 6개 부위로 구성된 구성요소인 머리, 몸통, 왼팔, 오른팔, 왼발, 오른발 등을 검출하여 추적한 후, 각 구성요소의 중심값과 색상정보를 이용하여 이전 프레임과 현재 프레임 간에 연결성을 두여 각 구성요소를 개별적으로 확률 전파를 통해 추적되어지고, 각 구성요소의 추적 결과는 구성요소들의 추정 결과를 구성요소 기반 확률 전파를 이용하여 인체의 동작을 추정하는 방법을 제안한다. 입력 영상에서 피부색 등의 색상 정보를 이용하여 인체 부위 또는 인체 모델의 구성 요소들 각각의 중심값과 색상정보를 가지고 확률전파를 통해 이것이 어떤 동작인지 동작 추정이 가능하다. 본 논문에서 제안하는 인체 동작 추적 시스템은 유아의 동작교육에 이용되는 7가지 동작인 걷기, 뛰기, 앙감질, 구부리기, 뻗기, 균형 잡기, 회전하기 등에 적용하였다. 본 논문에서 제안한 인체 모델의 각 구성요소 부위들을 독립적으로 검출하여 평균 96%의 높은 인식률을 나타냈고, 앞서 적용한 7가지 동작에 대해서 실험한 결과 평균 88.5% 성공률을 획득함으로써 본 논문에서 제안한 방법의 타당성을 보였다.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.26 no.1
• /
• pp.34-44
• /
• 1990

This paper describe on the target strengths for the dorsal and the side aspects of swimbladdered fishes, Mullet Mugil cephalus, Flounder Pleuronichthys cornutus and Rock trout Hexagrammos otakii two frequencies of 50KHz and 200KHz in the experimental water tank in order to improve the biomass estimation by the scientific fish finder. The results obtained are as follows: 1. The average of maximum target strength normalized by squared total length in cm unit are almost ranging from -70.9 dB to -66.8 dB regardless of species or frequencies. 2. The average of maximum target strength normalized by two-thirds squared body weight in g unit are almost ranging from -57.1 dB to -54.1 dB regardless of species or frequencies. 3. In comparison with target strength for 50KHz and 200KHz the former is more strong than the latter.

• The Journal of the Acoustical Society of Korea
• /
• v.25 no.6
• /
• pp.305-311
• /
• 2006

In this paper the performance of a array invariant method is evaluated for source-range estimation in horizontally stratified shallow water ocean waveguide. The method has advantage of little computationally effort over existing source-localization methods. such as matched field processing or the waveguide invariant and array gain is fully exploited. And. no knowledge of the environment is required except that the received field should not be dominated by purely interference This simple and instantaneous method is applied to simulated acoustic propagation filed for testing range estimation performance. The result of range estimation according to the SNR for the underwater impulsive source with broadband spectrum is demonstrated. The spatial smoothing method is applied to suppress the effect of mutipath propagation by high frequency signal. The result of performance test for range estimation shows that the error rate is within 20% at the SNR above 10dB.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.38 no.2
• /
• pp.119-128
• /
• 2002

This study investigates dorsal aspect target strength with fish size, tilt angle and frequency characteristics for the schlegel's black rockfish(Sebastes achlegeli) and the red seabream (Pagrus major). This study was carried out on free swimming fish in a cage in order to obtain acoustic data of the biomass estimation using the scientific echo sounder. The results obtained from this study are summarized as follows; 1 The coefficients of the schlegel's black rockfish and the red seabream using maximum TS with fish length were expressed -63.7dB and -62.6dB at a frequency of 38kHz, -64.4dB and -65.4dB at 120kHz, and -62.4dB and -65.0dB at 200kHz, respectively. 2. The coefficients of the schlegel\`s black rockfish and the red seabream using averaged TS with fish length were expressed -68.4dB and -67.9dB at a frequency of 38kHz, -73.4dB and -72.7dB at 120kHz, and -70.BdE and -73.4dB at 2001Hs, respectively. 3. The coefficients of the schlegel's black rockfish and the red seabream using maximum TS with body weight were expressed -52.0dB and -50.9dB at a frequency of 38kHz, -52.7dB and -53.7dB at 120kHz, and -50.7dB and -53.3dB at 200kHz, respectively. 4. The coefficients of the schlegel's black rockfish and the red seabream using averaged TS with body weight were expressed -56.7dB and -56.2dB at a frequency of 38kHz, -61.7dB and -61.0dB at 120kHz, and -59.ldE and -61.6dB at 200kHz, respectively. 5. Varying the tiIt angle of the two red seabream from -26$^{\circ}$to +25$^{\circ}$, the variation width of target strength expressed smaller at a frequency of 38kHz than at 120kHz and expressed about 3~6dB higher head up than head down at 120kHz.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.18 no.6
• /
• pp.1321-1330
• /
• 2023

This study attempts to address the problem of 3D pose estimation for multiple human objects through a single image generated during the character development process that can be used in augmented reality. In the existing top-down method, all objects in the image are first detected, and then each is reconstructed independently. The problem is that inconsistent results may occur due to overlap or depth order mismatch between the reconstructed objects. The goal of this study is to solve these problems and develop a single network that provides consistent 3D reconstruction of all humans in a scene. Integrating a human body model based on the SMPL parametric system into a top-down framework became an important choice. Through this, two types of collision loss based on distance field and loss that considers depth order were introduced. The first loss prevents overlap between reconstructed people, and the second loss adjusts the depth ordering of people to render occlusion inference and annotated instance segmentation consistently. This method allows depth information to be provided to the network without explicit 3D annotation of the image. Experimental results show that this study's methodology performs better than existing methods on standard 3D pose benchmarks, and the proposed losses enable more consistent reconstruction from natural images.

• Journal of the Korean Society of Fisheries and Ocean Technology
• /
• v.37 no.2
• /
• pp.133-139
• /
• 2001

In order to obtain fundamental data for estimation of fisheries resource by echo sounder, we carried out the measuring of dorsal aspect Target strengths for rock bream and dusky spinefoot fishes that were caught much around the Jeju Island and in South Sea of Korea. The appropriate equations share the common form. TS=A+20 log L, where TS is the average dorsal aspect target strength in decibels, L is the fish total length in centimeters, and the coefficient A is determined by a least mean squares regression analysis. For rock bream, the result is TS=-72.97+20 log L and, for dusky spinefoot it is TS=-63.16+20 log L And, we have investigated the bearing range of maximum dorsal aspect target strength for all of rock bream and dusky spinefoot by the echo sounder with transducer of which frequency is 200kHz. They are $-12^\circ$-$-21^\circ$and $-1^\circ$--8 espectively, when the fishes is swimming down to the bottom. The maximum dorsal target strengths are -41.50dB at -18 or rock bream and -30.69dB at $-6^\circ$for dusky spinefoot.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.48 no.6
• /
• pp.27-34
• /
• 2011

In this paper, we propose an active 3D shape acquisition method based on photometric stereo using camera and flash on a smartphone. Two smartphones are used as the master and slave, in which the slave projects illumination from different locations while the master captures the images and processes photometric stereo algorithm to reconstruct 3D shape. In order to reduce the error, the smartphone's camera is calibrated to overcome the effect of the lens distortion and nonlinear camera sensor response. We apply 5-point algorithm to estimate the pose between smartphone cameras and then estimate lighting direction vector to run the photometric stereo algorithm. Experimental result shows that the proposed system enables us to use smartphone as a 3D camera with low cost and high quality.

• Journal of Internet Computing and Services
• /
• v.18 no.6
• /
• pp.85-92
• /
• 2017

Recently, in the field of video surveillance, a Deep Learning based learning method has been applied to a method of detecting a moving person in a video and analyzing the behavior of a detected person. The human activity recognition, which is one of the fields this intelligent image analysis technology, detects the object and goes through the process of detecting the body keypoint to recognize the behavior of the detected object. In this paper, we propose a method for Body Keypoint Localization based on Object Detection using RGB-D information. First, the moving object is segmented and detected from the background using color information and depth information generated by the two cameras. The input image generated by rescaling the detected object region using RGB-D information is applied to Convolutional Pose Machines for one person's pose estimation. CPM are used to generate Belief Maps for 14 body parts per person and to detect body keypoints based on Belief Maps. This method provides an accurate region for objects to detect keypoints an can be extended from single Body Keypoint Localization to multiple Body Keypoint Localization through the integration of individual Body Keypoint Localization. In the future, it is possible to generate a model for human pose estimation using the detected keypoints and contribute to the field of human activity recognition.

• Journal of Navigation and Port Research
• /
• v.38 no.6
• /
• pp.585-591
• /
• 2014

The primary objective of the current work is to predict speed performance of the medium patrol boat over $F_N=0.5$ employing experimental materials based on the CFD before model tests. In other words, the predicted brake powers according to ship speeds are assessed satisfying the main engine capacity. The subject ships are selected the two different stern hull forms. The flow computation are conducted considering free surface and dynamic trim using a commercial CFD code(STAR-CCM+). The resistances of the bare-hull are obtained from CFD. Wave patterns, pressures and limiting streamlines on the hull and velocity distribution in the propeller plane for the two hull forms are compared using CFD. The effective powers of the object ships are assessed based on CFD. Resistance increase according to the attached appendages and quasi-propulsive efficiency are employed the experimental datas. Speed performance prediction method concerning high speed vessels like a medium patrol boat is developed employing CFD and experimental data.

• IEMEK Journal of Embedded Systems and Applications
• /
• v.19 no.1
• /
• pp.9-17
• /
• 2024

It is widely recognized that for 3D human pose estimation (HPE), dataset acquisition is expensive and the effectiveness of augmentation techniques of conventional visual recognition tasks is limited. We address these difficulties by presenting a simple but effective method that augments input images in terms of viewpoints when training a 3D human pose estimation (HPE) model. Our intuition is that meaningful variants of the input images for HPE could be obtained by viewing a human instance in the images from an arbitrary viewpoint different from that in the original images. The core idea is to synthesize new images that have self-occlusion and thus are difficult to predict at different viewpoints even with the same pose of the original example. We incorporate this idea into the training procedure of the 3D HPE model as an augmentation stage of the input samples. We show that a strategy for augmenting the synthesized example should be carefully designed in terms of the frequency of performing the augmentation and the selection of viewpoints for synthesizing the samples. To this end, we propose a new metric to measure the prediction difficulty of input images for 3D HPE in terms of the distance between corresponding keypoints on both sides of a human body. Extensive exploration of the space of augmentation probability choices and example selection according to the proposed distance metric leads to a performance gain of up to 6.2% on Human3.6M, the well-known pose estimation dataset.