• 한국방송∙미디어공학회:학술대회논문집
• /
• 한국방송∙미디어공학회 2020년도 추계학술대회
• /
• pp.252-255
• /
• 2020

Unsupervised deep learning methods have shown impressive results for the challenging monocular depth estimation task, a field of study that has gained attention in recent years. A common approach for this task is to train a deep convolutional neural network (DCNN) via an image synthesis sub-task, where additional views are utilized during training to minimize a photometric reconstruction error. Previous unsupervised depth estimation networks are trained within a fixed depth estimation range, irrespective of its possible range for a given image, leading to suboptimal estimates. To overcome this suboptimal limitation, we first propose an unsupervised adaptive depth estimation method guided by minimum and maximum (min-max) depth priors for a given input image. The incorporation of min-max depth priors can drastically reduce the depth estimation complexity and produce depth estimates with higher accuracy. Moreover, we propose a novel network architecture for adaptive depth estimation, called the AdaMM-DepthNet, which adopts the min-max depth estimation in its front side. Intensive experimental results demonstrate that the adaptive depth estimation can significantly boost up the accuracy with a fewer number of parameters over the conventional approaches with a fixed minimum and maximum depth range.

• 대한전자공학회:학술대회논문집
• /
• 대한전자공학회 2003년도 신호처리소사이어티 추계학술대회 논문집
• /
• pp.171-174
• /
• 2003

In this paper, we propose the method of stereo images composition using adaptive dense disparity estimation. For the correct composition of stereo image and 3D virtual object, we need correct marker position and depth information. The existing algorithms use position information of markers in stereo images for calculating depth of calibration object. But this depth information may be wrong in case of inaccurate marker tracking. Moreover in occlusion region, we can't know depth of 3D object, so we can't composite stereo images and 3D virtual object. In these reasons, the proposed algorithm uses adaptive dense disparity estimation for calculation of depth. The adaptive dense disparity estimation is the algorithm that use pixel-based disparity estimation and the search range is limited around calibration object.

• 반도체디스플레이기술학회지
• /
• 제14권4호
• /
• pp.92-97
• /
• 2015

The purpose of this work is to support a solution of optimizing TMO (tone mapping operator). JPEG XT Profile A and C utilize Erik Reinhard TMO that works well in most cases, however, detailed information of a scene is lost in some cases. Reinhard TMO only calculates its coefficient to have tone-mapping curve from log-average luminance, and this lead to lose details of bright and dark area of scenes in turn. Thus, this paper proposes an enhancement of the default TMO for JPEG XT Profile C to optimize tone-mapping curve. Main idea is that we divide tone mapping curve into several ranges, and set reasonable parameters for each range. By the experimental results, the proposed scheme shows and obtains better performance within a dark scene, compared to the default Reinhard TMO.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제9권3호
• /
• pp.1121-1139
• /
• 2015

In texture-plus-depth format, depth map compression is an important task. Different from normal texture images, depth maps have less texture information, while contain many homogeneous regions separated by sharp edges. This feature will be employed to form an efficient depth map coding scheme in this paper. Firstly, the histogram of the depth map will be analyzed to find an appropriate threshold that segments the depth map into the foreground and background regions, allowing the edge between these two kinds of regions to be obtained. Secondly, the two regions will be encoded through rate distortion optimization with a shape adaptive wavelet transform, while the edges are lossless encoded with JBIG2. Finally, a depth-updating algorithm based on the threshold and the depth range is applied to enhance the quality of the decoded depth maps. Experimental results demonstrate the effective performance on both the depth map quality and the synthesized view quality.

• International Journal of Fuzzy Logic and Intelligent Systems
• /
• 제16권2호
• /
• pp.131-139
• /
• 2016

Recognition of human motions has become a main area of computer vision due to its potential human-computer interface (HCI) and surveillance. Among those existing recognition techniques for human motions, head detection and tracking is basis for all human motion recognitions. Various approaches have been tried to detect and trace the position of human head in two-dimensional (2D) images precisely. However, it is still a challenging problem because the human appearance is too changeable by pose, and images are affected by illumination change. To enhance the performance of head detection and tracking, the real-time three-dimensional (3D) data acquisition sensors such as time-of-flight and Kinect depth sensor are recently used. In this paper, we propose an effective feature extraction method, called adaptive local binary pattern (ALBP), for depth image based applications. Contrasting to well-known conventional local binary pattern (LBP), the proposed ALBP cannot only extract shape information without texture in depth images, but also is invariant distance change in range images. We apply the proposed ALBP for head detection and tracking in depth images to show its effectiveness and its usefulness.

• 한국정밀공학회지
• /
• 제15권2호
• /
• pp.114-123
• /
• 1998

In the end milling process, the information of the surface errors plays an important role in adaptive control systems for precision machining. As the measuring accuracy of the surface errors directly matches the control's, it is an important factor for evaluating the performance of the system. In order to obtain the surface errors, the prediction using the cutting force, torque, motor power etc. is frequently practiced owing to the easiness in measurement. In the implementation of the prediction, the information on the cutting depths make it concrete and precise. Actually the axial depth of cut limits the range of the calculation. In general, it is not easy to know the cutting depths due to irregular shape of workpieces, inaccurate positioning of them on the table of machine tool, and machining error in the previous cutting. In addition to, even if cutting depths are informed, it is difficult to match the individual position of the cutter on the varying shape of the work material. This work suggests an algorithm estimating the cutting depths based on cutting force and makes it precise to predict the surface error. The proposed algorithm can be applied in more extensive cutting situations, such as presence of the tool wear, change of the work material hardness, etc.

• Smart Structures and Systems
• /
• 제16권3호
• /
• pp.497-519
• /
• 2015

This study discusses the use of Adaptive-Network-Based-Fuzzy-Inference-System (ANFIS) in predicting the shear strength of reinforced-concrete deep beams. 139 experimental data have been collected from renowned publications on simply supported high strength concrete deep beams. The results show that the ANFIS has strong potential as a feasible tool for predicting the shear strength of deep beams within the range of the considered input parameters. ANFIS's results are highly accurate, precise and therefore, more satisfactory. Based on the Sensitivity analysis, the shear span to depth ratio (a/d) and concrete cylinder strength ($f_c^{\prime}$) have major influence on the shear strength prediction of deep beams. The parametric study confirms the increase in shear strength of deep beams with an equal increase in the concrete strength and decrease in the shear span to-depth-ratio.

• 오토저널
• /
• 제14권6호
• /
• pp.113-126
• /
• 1992

Recently the multi-kind, small-amount manufacturing system has been replacing the mass manufacturing system, and domestic machining inustry also is eager to absorb the new technology because of its high productivity and cost reduction. The optimization of the cutting condition has been a vital problem in the machining industry, which would help increase the productivity and raise the international competitiveness. It is intended in this study to investigate the machining costs per unit time which is essential to the analysis of the optimal cutting condition, to computer the cutting speed that lead to the minimum machining costs and the maximum production to suggest the cutting speed range that enables efficient speed cutting, and to review the machining economy in relation to cutting depth and feed. Also considered are the optimal cutting speed and prodution rated in rrelation with feed. It is found that the minimum-cost cutting speed increases and the efficient cutting speed range is reduced as machining cost per unit time increases since the cutting speed for maximum production remains almost constant. The machining cost is also lowered and the production rate increases as the feed increases, and the feed should be selected to satisfy the required surface roughness. The machining cost and production rate are hardly affected by the cutting depth if the cutting speed stays below 100m/min, however, they are subject to change at larger cutting depth and the high-efficient speed range also is restricted. It can be established an adaptive optimal cutting conditions can be established in workshop by the auto-selection progam for optimal operation. It is expected that this method for choosing the optimal cutting conditions might contribute to the improvement of the productivity and reduced the cost. It is highly recommended to prepare the optimal cutting conditionthus obtained for future use in the programing of G-function of CNC machines. If proper programs that automatically select the optimal cutting conditions should be developed, it would be helpful to the works being done in the machine shops and would result in noticeable production raise and cost reduction.

• 방송공학회논문지
• /
• 제18권4호
• /
• pp.647-650
• /
• 2013

HEVC RExt(High Efficiency Video Coding Range Extension)는 RGB/YUV 4:2:2 4:4:4 색 샘플링 영상과 10비트 심도 이상의 영상 지원을 목표로 한다. RGB 영상은 YUV 4:2:0 색 샘플링 영상과는 달리 색평면 간 높은 상관도를 갖고 있으며, 이를 이용하여 화소값을 예측하는 기법들이 JCT-VC 표준화 회의에서 기고되었다. 하지만 일반적으로 RGB 영상의 고주파수 성분은 색평면 간 낮은 상관도를 갖고 있으며, 이는 색평면 간 예측 시 부호화 효율 저하의 원인이 된다. 따라서 본 논문에서는 색평면 간 예측 시 고주파수 성분을 저역통과필터를 통해 적응적으로 제거하는 기법을 제안한다. HEVC RExt의 RGB 영상을 통한 실험 결과, 본 논문에서 제안하는 기법은 기존 색평면 간 예측 기법에 비해 큰 복잡도의 증가 없이 평균 0.6%의 BD(Bjontegaard Distortion)-율 이득을 얻을 수 있었다.

• The Plant Pathology Journal
• /
• 제38권2호
• /
• pp.102-114
• /
• 2022

Pectobacterium carotovorum subsp. carotovorum (Pcc) is a gram-negative, broad host range bacterial pathogen which causes soft rot disease in potatoes as well as other vegetables worldwide. While Pectobacterium infection relies on the production of major cell wall degrading enzymes, other virulence factors and the mechanism of genetic adaptation of this pathogen is not yet clear. In the present study, we have performed an in-depth genome-wide characterization of Pcc strain ICMP5702 isolated from potato and compared it with other pathogenic bacteria from the Pectobacterium genus to identify key virulent determinants. The draft genome of Pcc ICMP5702 contains 4,774,457 bp with a G + C content of 51.90% and 4,520 open reading frames. Genome annotation revealed prominent genes encoding key virulence factors such as plant cell wall degrading enzymes, flagella-based motility, phage proteins, cell membrane structures, and secretion systems. Whereas, a majority of determinants were conserved among the Pectobacterium strains, few notable genes encoding AvrE-family type III secretion system effectors, pectate lyase and metalloprotease in addition to the CRISPR-Cas based adaptive immune system were uniquely represented. Overall, the information generated through this study will contribute to decipher the mechanism of infection and adaptive immunity in Pcc.