• 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.

• 전기전자학회논문지
• /
• 제22권1호
• /
• pp.68-73
• /
• 2018

본 논문에서는 $360^{\circ}$ 원형영상을 평면영상에 매핑하기 위한 효율적인 알고리즘을 제안한다. 제안된 알고리즘은 평면영상의 크기 구하기, 카메라와 평면영상 사이의 각 지점에 따른 거리 계산, 카메라와 평면영상이 수평으로 이루는 각도 산출, 카메라와 평면영상이 수직으로 이루는 각도 산출, $360^{\circ}$ 원형영상의 픽셀들을 평면영상의 픽셀에 매칭시키는 픽셀의 위치 계산 등 5단계로 구성된다. 제안한 $360^{\circ}$ 원형영상을 평면영상에 매핑하기 위한 효율적인 알고리즘을 평가하기 위하여 실험한 결과, 매핑된 평면영상의 복원율은 99%, 매핑된 평면영상의 화질은 72%로서 상용 소프트웨어의 기준치보다 높은 결과를 산출했기 때문에 알고리즘의 효용성이 확인되었다.

• 한국산학기술학회논문지
• /
• 제16권3호
• /
• pp.2192-2200
• /
• 2015

본 논문에서는 히스토그램 균등화 기반의 효율적인 차량용 영상 보정 알고리즘을 제안한다. 제안된 차량용 영상보정 알고리즘은 움직임 추정 및 움직임 보상을 통해 차량용 영상의 흔들림을 제거하였다. 그리고 영상을 보정하기 위해 영상을 일정 영역으로 분할하여 각각의 서브 영상에서 픽셀 값의 히스토그램을 계산하였다. 또한, 기울기를 조절하여 영상을 개선하였다. 제안된 알고리즘은 IP에 적용하여 성능 및 시간, 영상의 차이점을 평가하고, 차량용 카메라 영상의 흔들림 제거와 영상 개선을 확인하였다. 본 논문에서 제안된 차량용 영상 보정 알고리즘은 기존 차량 영상 안정화 기술과 비교하였을 때, 차량용 영상에 대한 흔들림 제거는 메모리를 사용하지 않고 실시간 처리를 했기 때문에 효율성을 입증하였다. 그리고 블록 정합을 통한 연산으로 계산 시간 감소 효과를 얻었고, 노이즈가 가장 적고 영상의 자연스러움이 더 뛰어난 복원 결과를 얻을 수 있었다.

• 방송공학회논문지
• /
• 제19권1호
• /
• pp.118-121
• /
• 2014

영상 보간법은 영상의 크기 변환에서 나타나는 새로운 좌표의 화소의 값을 결정하는 방법이다. 영상 콘텐츠가 대용량화되면서 고속으로 개선된 결과 영상을 생성할 수 있는 영상 보간법이 요구된다. 따라서 본 논문에서는 영상 해상도 개선을 위한 고속 다중 혼합영상 보간법을 제안한다. 제안하는 방법은 입력 영상의 4개의 부영상으로부터 12개의 예상되는 부족분을 추정하고, 추정된 부족분과 입력 영상을 결합한 후 보간 함수를 거쳐서 결과 영상을 생성한다. 제안하는 방법은 비교 방법들보다 PSNR에서 최대 1.9dB, SSIM에서 최대 0.052 개선된 결과를 나타내었으며, 주관적 화질 비교에서도 우위에 있음을 실험을 통해서 알 수 있었다. 알고리즘의 동작속도 비교를 통해서 기존의 방법들보다 최소 3배 이상 빠르게 동작하는 것을 알 수 있었다. 제안하는 방법은 영상 해상도 개선을 위한 응용 환경에서 유용하게 사용될 수 있다.

• 한국문화재보존과학회:학술대회논문집
• /
• 한국문화재보존과학회 2006년도 제24회 학술대회발표 논문집
• /
• pp.11-15
• /
• 2006

This work aims to clarify the character and selection of repair papers, based on fibre orientation information of paper cultural properties under restoration; using nondestructive image analysis applied to micrographs of paper surfaces. Image analysis using fast Fourier transform with suitable modifications was demonstrated to be an effective means to determine angle and intensity of fiber orientation as a nondestructive method. In the case of contemporary repair papers, the Japanese handmade paper exhibited a high anisotropy value, while the Korean handmade papers exhibited low anisotropy values. Korea and Japanese traditional hand making papers were well distinguished. Korea and Japanese papers made in the traditional ways showed its own characteristic orientation behavior in accordance with the motion of a bamboo wire.

• Journal of applied mathematics & informatics
• /
• 제31권3_4호
• /
• pp.353-363
• /
• 2013

A variant of the global conjugate gradient method for solving general linear systems with multiple right-hand sides is proposed. This method is called as the global conjugate gradient linear least squares (Gl-CGLS) method since it is based on the conjugate gradient least squares method(CGLS). We present how this method can be implemented for the image deblurring problems with Neumann boundary conditions. Numerical experiments are tested on some blurred images for the purpose of comparing the computational efficiencies of Gl-CGLS with CGLS and Gl-LSQR. The results show that Gl-CGLS method is numerically more efficient than others for the ill-posed problems.

• 한국군사과학기술학회지
• /
• 제10권1호
• /
• pp.101-106
• /
• 2007

In this paper, we propose chain code based decision technique of edge orientation. Edge Information is one of the most important information for handling image signals, and is applied to various civilian and military area, such as image surveillance for military reconnaissance, restoration of noised image signal, fingerprint recognition, and so on. The conventional methods to find edge orientations exploit a certain size of mask which weights for a certain direction, so they showed limitations in the case of diagonal edges except 45 degrees. We find edge orientations using chain code. According to the experiments, the proposed method shows better performance than the conventional approaches. The exact edge orientations including diagonal directions are well found.

• 대한의용생체공학회:의공학회지
• /
• 제10권3호
• /
• pp.293-302
• /
• 1989

A Dual Sensor Wiener Filter technique was used to improve the image quality of the scanning type digital radiographic system (resolution and SNR). In this method, two images were acquried simultaneously using two sensors with high and low resolution and SNR values. Using the cross Power spectrum between dual sensor outputs of the same chest radiographic image. we design a new type of Wiener Filter and implement it with fast algorithm. We compared the performance of this new dual sensor filter with conventional single sensor filters (Wiener Filter and Parametric Projec- tion Filter) . In simulation studies, it is shown that this new method has SNR improvement of 1-3 dB better than conventional filters.

• 대한임베디드공학회논문지
• /
• 제10권1호
• /
• pp.1-6
• /
• 2015

Recently, handy pulse measurement method was introduced by using smart phone camera. However, measured values are not consistent with the variations of external light conditions, because the external light interfere with dynamic range of captured pulse image. Thus, adaptive dynamic range reconstruction algorithm is proposed to conduct pulse measurement robust to light condition. The minimum and maximum values for dynamic ranges of green and blue channels are adjusted to appropriate values for pulse measurement. In addition, sigmoid function based curve is applied to adjusted dynamic range. Experimental results show that the proposed algorithm conducts suitably dynamic range reconstruction of pulse image for the interference of external light sources.

• 대한전자공학회논문지
• /
• 제26권8호
• /
• pp.1238-1246
• /
• 1989

In this paper, a texture synthesis technique based on the NCAR(non-causal auto-regressive) model and the pyramid structure is proposed. In order to estimate the NCAR model parameters accurately from a noisy texture, the MAP(maximum a posteriori) estimation technique is also employed. In our approach, since the input texture is decomposed into the Laplacian oyramid planes first and then the NCAR model is applied to each plane, we are able to obtain a good synthesized texture even if the texture exhibits some non-random local structure or non-homogenity. The usrfulness of the proposed method is demonstrated with seveal real textures in the Brodatz album. Finally, the 2-dimensional MAP estimation technique can be used to the image restoration for noisy images as well as a texture image synthesis.