• Proceedings of the Korea Information Processing Society Conference
• /
• 2022.11a
• /
• pp.646-647
• /
• 2022

Multiview stereo (MVS) 3D reconstruction of a scene from images is a fundamental computer vision problem that has been thoroughly researched in recent times. Traditionally, MVS approaches create dense correspondences by constructing regularizations and hand-crafted similarity metrics. Although these techniques have achieved excellent results in the best Lambertian conditions, traditional MVS algorithms still contain a lot of artifacts. Therefore, in this study, we suggest using a transformer network to accelerate the MVS reconstruction. The network is based on a transformer model and can extract dense features with 3D consistency and global context, which are necessary to provide accurate matching for MVS.

• Proceedings of the Optical Society of Korea Conference
• /
• 2008.02a
• /
• pp.113-114
• /
• 2008

• Journal of Information Processing Systems
• /
• v.17 no.6
• /
• pp.1083-1096
• /
• 2021

Compressed sensing-based matching pursuit algorithms can estimate the sparse channel of massive multiple input multiple-output systems with short pilot sequences. Although they have the advantages of low computational complexity and low pilot overhead, their accuracy remains insufficient. Simply multiplying the weight value and the estimated channel obtained in different iterations can only improve the accuracy of channel estimation under conditions of low signal-to-noise ratio (SNR), whereas it degrades accuracy under conditions of high SNR. To address this issue, an improved weighted matching pursuit algorithm is proposed, which obtains a suitable weight value u_op by training the channel data. The step of the weight value increasing with successive iterations is calculated according to the sparsity of the channel and u_op. Adjusting the weight value adaptively over the iterations can further improve the accuracy of estimation. The results of simulations conducted to evaluate the proposed algorithm show that it exhibits improved performance in terms of accuracy compared to previous methods under conditions of both high and low SNR.

• Smart Structures and Systems
• /
• v.25 no.2
• /
• pp.123-133
• /
• 2020

Compressive sensing (CS) is a newly developed data acquisition and processing technique that takes advantage of the sparse structure in signals. Normally signals in their primitive space or format are reconstructed from their compressed measurements for further treatments, such as modal analysis for vibration data. This approach causes problems such as leakage, loss of fidelity, etc., and the computation of reconstruction itself is costly as well. Therefore, it is appealing to directly work on the compressed data without prior reconstruction of the original data. In this paper, a direct approach for modal analysis of damped systems is proposed by decomposing the compressed measurements with an appropriate dictionary. The damped free vibration function is adopted to form atoms in the dictionary for the following sparse decomposition. Compared with the normally used Fourier bases, the damped free vibration function spans a space with both the frequency and damping as the control variables. In order to efficiently search the enormous two-dimension dictionary with frequency and damping as variables, a two-step strategy is implemented combined with the Orthogonal Matching Pursuit (OMP) to determine the optimal atom in the dictionary, which greatly reduces the computation of the sparse decomposition. The performance of the proposed method is demonstrated by a numerical and an experimental example, and advantages of the method are revealed by comparison with another such kind method using POD technique.

• Proceedings of the KIEE Conference
• /
• 2006.10c
• /
• pp.536-538
• /
• 2006

The most important factor of mobile robot is to build a map for surrounding environment and estimate its localization. This paper proposes a real-time localization and map building method through 3-D reconstruction using scale invariant feature from monocular camera. Mobile robot attached monocular camera looking wall extracts scale invariant features in each image using SIFT(Scale Invariant Feature Transform) as it follows wall. Matching is carried out by the extracted features and matching feature map that is transformed into absolute coordinates using 3-D reconstruction of point and geometrical analysis of surrounding environment build, and store it map database. After finished feature map building, the robot finds some points matched with previous feature map and find its pose by affine parameter in real time. Position error of the proposed method was maximum. 8cm and angle error was within $10^{\circ}$.

• Proceedings of the Korean Society of Computer Information Conference
• /
• 2018.07a
• /
• pp.389-390
• /
• 2018

본 논문에서는 3차원 복원을 위한 특징점 추출 및 매칭에 대한 보다 정확한 방법을 제안한다. 이 방법은 컴퓨터 비전의 기본이 되는 분야로 복원뿐 만 아니라 SLAM과 같은 지도 작성 및 자율 운행에도 필요한 방법이다. 본 연구는 3차원 물체 복원을 위해서 사용하는 방법 중 하나인 Column space fitting(CSF)을 이용하여 turntable-image data에 적용하여 성능을 평가하여 정확성을 검증을 한다. 오늘날 3D scanner를 이용하여 물체를 3차원 모델을 획득하고 3D프린터를 이용하여 다양한 분야에 적용한다. 그러나 고가의 장비이기 때문에 접근성이 떨어진다. 본 연구는 영상들만을 가지고 기하학적 계산을 통해 3차원 모델을 획득한다. 본 연구결과는 기존의 방법인 KLT 알고리즘과 비교하여 RMSE의 값을 약 5배를 줄이는 성능 향상을 보인다.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.07a
• /
• pp.144-145
• /
• 2020

본 논문에서는 정밀한 3차원 복원 및 시점 합성을 위해 매칭 비용을 반복적으로 업데이트하는 Generalized Soft 3D Reconstruction (GenSoft3D) 알고리즘을 제안한다. 먼저 다시점 영상들과 카메라 자세정보가 주어지면 GenSoft3D는 볼륨 기반의 다시점 스테레오 매칭 알고리즘으로 시점별 초기 매칭 비용 볼륨 및 시차 맵을 계산한다. 그 후 정제 과정에서 각 시점은 모든 시차 맵을 이용하여 표면 확률 및 가시 확률을 계산한다. 표면 확률은 초기 매칭 비용 업데이트에 사용하며, 가시 확률은 폐색 영역의 정확한 시차를 계산하기 위해 사용된다. 해당 정제 과정을 일정 횟수 반복할 경우 시점별 고정밀의 시차 맵 획득이 가능하다. 또한 시차 맵의 정확도가 향상됨에 따라 정확한 시점 합성이 가능하다.

• The KIPS Transactions:PartB
• /
• v.17B no.4
• /
• pp.295-302
• /
• 2010

Example-based super resolution(EBSR) is a method to reconstruct high-resolution images by learning patch-wise correspondence between high-resolution and low-resolution images. It can reconstruct a high-resolution from just a single low-resolution image. However, when it is applied to a text image whose font type and size are different from those of training images, it often produces lots of noise. The primary reason is that, in the patch matching step of the reconstruction process, input patches can be inappropriately matched to the high-resolution patches in the patch dictionary. In this paper, we propose a new patch matching method to overcome this problem. Using an image observation model, it preserves the correlation between the input and the output images. Therefore, it effectively suppresses spurious noise caused by inappropriately matched patches. This does not only improve the quality of the output image but also allows the system to use a huge dictionary containing a variety of font types and sizes, which significantly improves the adaptability to variation in font type and size. In experiments, the proposed method outperformed conventional methods in reconstruction of multi-font and multi-size images. Moreover, it improved recognition performance from 88.58% to 93.54%, which confirms the practical effect of the proposed method on recognition performance.

• Progress in Medical Physics
• /
• v.20 no.2
• /
• pp.62-71
• /
• 2009

Recently, optical coherence tomography (OCT) has demonstrated considerable promise for the noninvasive assessment of biological tissues. However, OCT images difficult to analyze due to speckle noise. In this paper, we tested various image processing techniques for speckle removal of human and rabbit cartilage OCT images. Also, we distinguished the images which get with methods of image segmentation for OCT images, and found the most suitable method for segmenting an image. And, we selected image segmentation suitable for OCT before image reconstruction. OCT was a weak point to system design and image processing. It was a limit owing to measure small a distance and depth size. So, good edge matching algorithms are important for image reconstruction. This paper presents such an algorithm, the chamfer matching algorithm. It is made of background for 3D image reconstruction. The purpose of this paper is to describe good image processing techniques for speckle removal, image segmentation, and the 3D reconstruction of cartilage OCT images.

• Archives of Plastic Surgery
• /
• v.37 no.4
• /
• pp.492-495
• /
• 2010

Purpose: Facial tumor excision is a common cause of lower eyelid defect in old patients. Many methods have been introduced for the reconstruction of lower eyelid. However, conventional surgical method can cause various complications like scar, ectropion and unnatural color matching. Thus, we introduce a simple and aesthetically acceptable method for the reconstruction of lower eyelid defect. Methods: Three elderly patients with skin cancer in the unilateral lower eyelid were operated by the new method. Following a wide excision of skin cancer, subcilliary incision of lower blepharoplasty was carried out. Elevated skin flap of lower eyelid was redrapped for the correction of defect and the remnant skin from lateral portion of lower eyelid was used for full thickness skin graft (FTSG) to correct the remaining defect. Results: All grafts survived and color match of the graft was excellent without ectropion. Furthermore, wrinkles of the lower eyelid were improved after the blepharoplasty. Conclusion: Lower eyelid defect resulting from wide excision of malignant tumor in old patients could be reconstructed successfully by modifying the conventional lower eyelid blepharoplasty along with FTSG using the remnant skin.