• Journal of the Korean Society for Industrial and Applied Mathematics
• /
• v.17 no.3
• /
• pp.197-207
• /
• 2013

The Cahn-Hilliard equation was proposed as a phenomenological model for describing the process of phase separation of a binary alloy. The equation has been applied to many physical applications such as amorphological instability caused by elastic non-equilibrium, image inpainting, two- and three-phase fluid flow, phase separation, flow visualization and the formation of the quantum dots. To solve the Cahn-Hillard equation, many numerical methods have been proposed such as the explicit Euler's, the implicit Euler's, the Crank-Nicolson, the semi-implicit Euler's, the linearly stabilized splitting and the non-linearly stabilized splitting schemes. In this paper, we investigate each scheme in finite-difference schemes by comparing their performances, especially stability and efficiency. Except the explicit Euler's method, we use the fast solver which is called a multigrid method. Our numerical investigation shows that the linearly stabilized stabilized splitting scheme is not unconditionally gradient stable in time unlike the known result. And the Crank-Nicolson scheme is accurate but unstable in time, whereas the non-linearly stabilized splitting scheme has advantage over other schemes on the time step restriction.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• fall
• /
• pp.139-143
• /
• 2021

본 논문에서는 꾸준히 연구되어 오던 이미지 복원 문제에서 초해상화와 인페인팅이라는 복합적 이미지 복원을 동시에 처리하는 해결 방법을 제안한다. 초해상화는 국지적 픽셀 정보를 이용하여 고해상도의 영상을 복원하고, 인페인팅은 이미지 전체 정보를 활용하여 영상 내 비어 있는 영역을 생성해야 하므로, 이러한 두 가지 영상 복원 기법을 동시에 수행하는 것은 상당히 어려운 문제이다. 그렇기에 인페인팅과 초해상화는 이미지 복원에서 널리 활용되는 기술인 만큼 동시에 해결할 수 있는 기법에 대한 수요는 있음에도 지금까지 거의 연구되지 않았다. 본 논문은 초해상화 및 인페인팅 합동 처리에 있어 복합적인 정보를 모두 다뤄야하는 네트워크가 서로의 성능을 저하시키지 않도록 개략적 복원 네트워크 (Coarse network), 디테일 복원 네트워크 (Refinement network), 초해상화 네트워크 (SR network)로 분리하여 초해상화 및 인페인팅 합동 처리를 수행하며, 각 단계마다 결과 영상을 얻어 스케일 별 정답 영상과 손실함수를 계산하여 복합적인 성능을 올릴 수 있는 방법을 제시한다. 또한 순차적 단일 모델에 비하여 인페인팅과 초해상화를 합동 학습하는 제안 모델이 개선된 화질의 결과 영상을 획득할 수 있다는 것을 실험적으로 보인다.

• Smart Media Journal
• /
• v.3 no.4
• /
• pp.28-34
• /
• 2014

Recently the depth-image-based rendering (DIBR) method is usually used in 3D image application filed. Virtual view image is created by using a known view with associated depth map to make a virtual view point which did not taken by the camera. But, disocclusion area occur because the virtual view point is created using a depth image based image 3D warping. To remove those kind of disocclusion region, many hole filling methods are proposed until now. Constant color region searching, horizontal interpolation, horizontal extrapolation, and variational inpainting techniques are proposed as a hole filling methods. But when using those hole filling method some problem occurred. The different types of annoying artifacts are appear in texture region hole filling procedure. In this paper to solve those problem, the multi-directional extrapolation method is newly proposed for efficiency of expanded hole filling performance. The proposed method is efficient when performing hole filling which complex texture background region. Consideration of directionality for hole filling method use the hole neighbor texture pixel value when estimate the hole pixel value. We can check the proposed hole filling method can more efficiently fill the hole region which generated by virtual view synthesis result.

• Journal of the Korea Computer Graphics Society
• /
• v.24 no.2
• /
• pp.1-9
• /
• 2018

In this paper, we propose a method to restore corrupted black and white facial images to color. Previous studies have shown that when coloring damaged black and white photographs, such as old ID photographs, the area around the damaged area is often incorrectly colored. To solve this problem, this paper proposes a method of restoring the damaged area of input photo first and then performing colorization based on the result. The proposed method consists of two steps: BEGAN (Boundary Equivalent Generative Adversarial Networks) model based restoration and CNN (Convolutional Neural Network) based coloring. Our method uses the BEGAN model, which enables a clearer and higher resolution image restoration than the existing methods using the DCGAN (Deep Convolutional Generative Adversarial Networks) model for image restoration, and performs colorization based on the restored black and white image. Finally, we confirmed that the experimental results of various types of facial images and masks can show realistic color restoration results in many cases compared with the previous studies.