• International Journal of Contents
• /
• 제7권4호
• /
• pp.19-23
• /
• 2011

This paper proposes a segmentation method for overlapped vehicles based on analysis of the vehicle location and the spatiotemporal association information. This method can be used in an intelligent transport system. In the proposed method, occlusion is detected by analyzing the association information based on a vehicle's location in continuous images, and occlusion segmentation is carried out by using the vehicle information prior to occlusion. In addition, the size variations of the vehicle to which association tracking is applied can be anticipated by learning the variations according to the overlapped vehicles' movements. To assess the performance of the suggested method, image data collected from CCTVs recording traffic information is used, and average success rate of occlusion segmentation is 96.9%.

• 대한인간공학회지
• /
• 제28권3호
• /
• pp.49-54
• /
• 2009

Visual occlusion method is a visual demand measuring technique which uses periodic vision/occlusion cycle to simulate driving environment. It became one of the most popular techniques for the evaluation of in-vehicle interfaces due to its robustness and cost-effectiveness. However, it has a limitation in that the vision/occlusion cycle forces the user to use the IVIS at a predetermined pace, while a driver decides when to use the device on his/her own in actual driving. This paper proposes a user-driven visual occlusion method for measuring the visual demand of in-vehicle interfaces. An experiment was conducted to examine the visual demand of an in-vehicle interface prototype using both the existing (system-driven) occlusion method and the proposed (user-driven) one. Two in-vehicle tasks were evaluated: address input and radio tuning. The results showed that, for the radio tuning task, there were significant differences in total shutter open time and resumability ratio between the methods. The user-driven visual occlusion method not only allows a better representation of drivers' behavior, but it also seems to provide more information on the chunkability of a task.

• 전자공학회논문지
• /
• 제54권2호
• /
• pp.99-105
• /
• 2017

스테레오 정합은 두 영상의 색상과 공간 유사성이 최대가 되는 지점을 찾아 깊이 정보를 예측한다. 그런데, 두 시점 사이에 발생하는 폐색 영역으로 잘못된 깊이 정보를 얻게 되고, 폐색 영역을 보완하지 않은 깊이 정보는 시간 축 스테레오 매칭에서 잡음을 전파하는 문제를 일으킨다. 본 논문은 폐색 영역을 보완하여 시간 축 상에서 발생하는 잡음의 전파를 줄이고, 정확한 깊이 정보를 공유하여 스테레오 매칭의 정확성을 높이는 방법을 제안한다. 제안한 알고리즘을 수행하기 위해 색상과 공간의 유사성을 계산하는 함수를 정의하여 초기 깊이 정보를 예측하고, 세 가지 제약사항을 고려한 에너지 함수를 세워 (EM: expectation maximization) 으로 폐색 영역을 구한 뒤, 동적 프로그래밍 방법으로 예측된 폐색 영역을 보정한다. 끝으로, 이전시점에 구해진 정확한 깊이 정보를 사용하여 시간 축 스테레오 매칭을 수행한다. 실험을 통해 제안한 알고리즘이 기존의 폐색영역 예측 방법보다 우수한 성능을 가지는 것을 알 수 있었으며, 시간 축 정보를 고려하지 않은 스테레오 매칭 방법보다 정확한 결과를 얻는 것을 확인할 수 있었다.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제15권7호
• /
• pp.2547-2567
• /
• 2021

It has been widely acknowledged that occlusion impairments adversely distress many face recognition algorithms' performance. Therefore, it is crucial to solving the problem of face image occlusion in face recognition. To solve the image occlusion problem in face recognition, this paper aims to automatically de-occlude the human face majority or discriminative regions to improve face recognition performance. To achieve this, we decompose the generative process into two key stages and employ a separate generative adversarial network (GAN)-based network in both stages. The first stage generates an initial coarse face image without an occlusion mask. The second stage refines the result from the first stage by forcing it closer to real face images or ground truth. To increase the performance and minimize the artifacts in the generated result, a new refine loss (e.g., reconstruction loss, perceptual loss, and adversarial loss) is used to determine all differences between the generated de-occluded face image and ground truth. Furthermore, we build occluded face images and corresponding occlusion-free face images dataset. We trained our model on this new dataset and later tested it on real-world face images. The experiment results (qualitative and quantitative) and the comparative study confirm the robustness and effectiveness of the proposed work in removing challenging occlusion masks with various structures, sizes, shapes, types, and positions.

• 대한인간공학회지
• /
• 제30권5호
• /
• pp.589-595
• /
• 2011

Objective: This study compares two visual occlusion methods for the evaluation of in-vehicle interfaces. Background: Visual occlusion is a visual demand measuring technique which uses periodic vision/occlusion cycle to simulate a driving(or mobile) environment. It has been widely used for the evaluation of in-vehicle interfaces. There are two major implementation methods for this technique: (1) occlusion using PLATO(portable liquid crystal apparatus for tachistoscopic occlusion) goggles; (2) occlusion using a software application on a touchscreen device. Method: An experiment was conducted to examine the visual demand of an in-vehicle interface prototype using the goggle-based and the touchscreen-based occlusion methods. Address input and radio tuning tasks were evaluated in the experiment. Results: The results showed that, for the radio tuning task, there were no significant differences in total shutter open time and resumability ratio between the two occlusionconditions. However, it took longer for the participants to input addresses with the touchscreen-based occlusion. Conclusion & Application: The results suggest that touchscreen-based method could be used as an alternative to traditional, gogglebased visual occlusion especially in less demanding visual tasks such as radio tuning.

• 한국멀티미디어학회논문지
• /
• 제18권2호
• /
• pp.158-167
• /
• 2015

This study presents a new method to improve the speed of high quality volume rendering. We improve the speed of ambient occlusion which is one of the global illumination techniques used in traditional volume visualization. Calculating ambient occlusion takes much time because it determines an illumination value of a sample by integrating opacities of nearby samples. This study proposes an improved method for this by using local statistics such as averages and standard deviations. We calculate local statistics for each volume block, a set of nearby samples, in pre-processing time. In the rendering process, we efficiently determine the illumination value by assuming the density distribution as a normal distribution. As the results, we can generate high quality images that combine ambient occlusion illumination with local illumination in real time.

• 정보처리학회논문지A
• /
• 제15A권1호
• /
• pp.1-8
• /
• 2008

차폐 선별 기법은 가시성 선별 기법 중 하나로, 다른 물체에 가려서 보이지 않는 물체나 삼각형에 대한 연산을 제외시키는 기법이다. 이는 불필요한 연산량을 효과적으로 줄이기 ??문에 복잡한 장면을 실시간으로 처리하기 위해 필수적이다. 하지만 기존의 차폐 선별 기법인 차폐 쿼리는 가시성 검사를 위해 물체 데이터를 하드웨어에 두 번 보내야 하며, 이로 인해 불필요한 연산이 발생한다. 또 다른 기존 하드웨어 차폐 선별 기법인 VCBP는 빠른 수행을 하지만 바운딩 볼륨의 검사를 지원하지 않으며 응용으로 그 결과를 보내는 기능이 없다. 본 논문에서는 이러한 문제점들을 해결한 가시성 선별과 렌더링을 한 번에 처리할 수 있는 단일 패스 알고리즘을 제안한다. 제안하는 기법은 일차적으로 3차원 가속 하드웨어의 초기 단계인 삼각형을 픽셀로 나누는 래스터화 단계에서 캐쉬를 이용하여 빠르게 가시성 선별을 수행한다. 그와 동시에 가시성 선별 과정에서는 각 프리미티브의 가시성 정보를 응용단계로 보낸다. 응용단계에서는 하드웨어로부터 받은 이전 프레임의 가시성 정보와 공간계층 트리 구조를 이용하여 하드웨어로 보내는 보이지 않는 프리미티브를 위한 데이터량을 획기적으로 줄인다. 제안하는 구조는 하드웨어 차폐 선별 쿼리를 이용하는 기존 이중 패스 알고리즘 중 S&W 대비 최대 44%, 최저 14%의 성능이 향상되었고, CHC 대비 최대 25%, 최저 17%의 성능이 향상되었다.

• Journal of Multimedia Information System
• /
• 제5권1호
• /
• pp.35-42
• /
• 2018

Direct volume rendering (DVR) is an important 3D visualization method for medical images as it depicts the full volumetric data. However, because DVR renders the whole volume, regions of interests (ROIs) such as a tumor that are embedded within the volume maybe occluded from view. Thus, conventional 2D cross-sectional views are still widely used, while the advantages of the DVR are often neglected. In this study, we propose a new visualization algorithm where we augment the 2D slice of interest (SOI) from an image volume with volumetric information derived from the DVR of the same volume. Our occlusion-based DVR augmentation for SOI (ODAS) uses the occlusion information derived from the voxels in front of the SOI to calculate a depth parameter that controls the amount of DVR visibility which is used to provide 3D spatial cues while not impairing the visibility of the SOI. We outline the capabilities of our ODAS and through a variety of computer tomography (CT) medical image examples, compare it to a conventional fusion of the SOI and the clipped DVR.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• 제13권4호
• /
• pp.2094-2112
• /
• 2019

Focusing on the issue that the conventional Kernel Correlation Filter (KCF) algorithm has poor performance in handling scale change and obscured objects, this paper proposes an anti-occlusion and scale adaptive tracking algorithm in the basis of KCF. The average Peak-to Correlation Energy and the peak value of correlation filtering response are used as the confidence indexes to determine whether the target is obscured. In the case of non-occlusion, we modify the searching scheme of the KCF. Instead of searching for a target with a fixed sample size, we search for the target area with multiple scales and then resize it into the sample size to compare with the learnt model. The scale factor with the maximum filter response is the best target scaling and is updated as the optimal scale for the following tracking. Once occlusion is detected, the model updating and scale updating are stopped. Experiments have been conducted on the OTB benchmark video sequences for compassion with other state-of-the-art tracking methods. The results demonstrate the proposed method can effectively improve the tracking success rate and the accuracy in the cases of scale change and occlusion, and meanwhile ensure a real-time performance.

• 국제학술발표논문집
• /
• The 9th International Conference on Construction Engineering and Project Management
• /
• pp.904-911
• /
• 2022

Occlusion is one of the most challenging problems for computer vision-based construction monitoring. Due to the intrinsic dynamics of construction scenes, vision-based technologies inevitably suffer from occlusions. Previous researchers have proposed the occlusion handling methods by leveraging the prior information from the sequential images. However, these methods cannot be employed for construction object detection in non-sequential images. As an alternative occlusion handling method, this study proposes a data augmentation-based framework that can enhance the detection performance under occlusions. The proposed approach is specially designed for rebar occlusions, the distinctive type of occlusions frequently happen during construction worker detection. In the proposed method, the artificial rebars are synthetically generated to emulate possible rebar occlusions in construction sites. In this regard, the proposed method enables the model to train a variety of occluded images, thereby improving the detection performance without requiring sequential information. The effectiveness of the proposed method is validated by showing that the proposed method outperforms the baseline model without augmentation. The outcomes demonstrate the great potential of the data augmentation techniques for occlusion handling that can be readily applied to typical object detectors without changing their model architecture.