• KIPS Transactions on Software and Data Engineering
• /
• v.11 no.11
• /
• pp.465-472
• /
• 2022

In this paper, the style synthesis network is trained to generate style-synthesized video through the style synthesis through training Stylegan and the video synthesis network for video synthesis. In order to improve the point that the gaze or expression does not transfer stably, 3D face restoration technology is applied to control important features such as the pose, gaze, and expression of the head using 3D face information. In addition, by training the discriminators for the dynamics, mouth shape, image, and gaze of the Head2head network, it is possible to create a stable style synthesis video that maintains more probabilities and consistency. Using the FaceForensic dataset and the MetFace dataset, it was confirmed that the performance was increased by converting one video into another video while maintaining the consistent movement of the target face, and generating natural data through video synthesis using 3D face information from the source video's face.

• Journal of Broadcast Engineering
• /
• v.11 no.4 s.33
• /
• pp.507-520
• /
• 2006

In this paper, we propose an application of OpenGL functions for novel view synthesis from multi-view images and depth maps. While image based rendering has been meant to generate synthetic images by processing the camera view with a graphic engine, little has been known about how to apply the given images and depth information to the graphic engine and render the scene. This paper presents an efficient way of constructing a 3D space with camera parameters, reconstructing the 3D scene with color and depth images, and synthesizing virtual views in real-time as well as their depth images.

• Proceedings of the Korean Institute of Intelligent Systems Conference
• /
• 2003.09a
• /
• pp.46-49
• /
• 2003

A concept of fuzzy wavelets is proposed by a fuzzification of morphological wavelets. In the proposed fuzzy wavelets, analysis and synthesis schemes can be formulated as the operations of fuzzy relational calculus. In order to perform an efficient compression and reconstruction, an alphaband is also proposed as a soft thresholding of the wavelets. In the image compression/reconstruction experiment using test images extracted Standard Image DataBAse (SIDBA), it is confirmed that the root mean square error (RMSE) of the proposed soft thresholding is decreased to 87.3％ of the conventional hard thresholding.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2021.06a
• /
• pp.105-108
• /
• 2021

Generative adversarial networks (GANs) have reached a great result at creating the synthesis image, especially in the face generation task. Unlike other deep learning tasks, the input of GANs is usually the random vector sampled by a probability distribution, which leads to unstable training and unpredictable output. One way to solve those problems is to employ the label condition in both the generator and discriminator. CelebA and FFHQ are the two most famous datasets for face image generation. While CelebA contains attribute annotations for more than 200,000 images, FFHQ does not have attribute annotations. Thus, in this work, we introduce a method to learn the attributes from CelebA then predict both soft and hard labels for FFHQ. The evaluated result from our model achieves 0.7611 points of the metric is the area under the receiver operating characteristic curve.

• Journal of the Korean Graphic Arts Communication Society
• /
• v.24 no.2
• /
• pp.49-59
• /
• 2006

In this paper, a transparent film exposed the effect of heat cut-off, reveal as means of the prevention to wrong operation of parts of display and forgery of the credit card, also it will intercept rising of the temperature in interior of a room and car by diminish the influx of near-infrared ray wavelength of solar energy come from the window. As in the past a film which absorb a wavelength of $800{\sim}2500nm$ in near-infrared ray, manufactured in physical vapor deposition(PVD), chemical vapor deposition(CVD) to using ATO, ITO of inorganic materials or sputtering method. but it has lots of problem in manufacture. On the other hand, recently a paper said it easily form a transparent film to using organic dye. This paper show synthesis of many derivatives used in Ni-complex and then it investigate to optical property and durability of flim by make the transparent film.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.8
• /
• pp.1136-1141
• /
• 2022

Researches of face recognition on masked faces have been increasingly important due to the COVID-19 pandemic. To realize a stable and practical recognition performance, large amount of facial image data should be acquired for the purpose of training. However, it is difficult for the researchers to obtain masked face images for each human subject. This paper proposes a novel method to synthesize a face image and a virtual mask pattern. In this method, a pair of masked face image and unmasked face image, that are from a single human subject, is fed into a convolutional autoencoder as training data. This allows learning the geometric relationship between face and mask. In the inference step, for a unseen face image, the learned convolutional autoencoder generates a synthetic face image with a mask pattern. The proposed method is able to rapidly generate realistic masked face images. Also, it could be practical when compared to methods which rely on facial feature point detection.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2020.11a
• /
• 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.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2003.11a
• /
• pp.153-156
• /
• 2003

As there is a growing tendency to represent the 3D content instead of the 2D content, researches for the stereoscopic image and video are under way in a variety of fields such as acquisition compression, transmission, authoring and display. The authoring technique for stereoscopic contents has given emphasis to virtual stereoscopic contents. Thus the authoring technique for stereoscopic pictures is insufficient. When we compose a stereo scene with stereoscopic pictures, stereoscopic contents may not match the stereo scene because each stereoscopic picture may have different camera condition. To solve this problem, stereoscopic pictures have been modified manually. It is a laborious work and will be spent much time. Also it is difficult for a user who does not have an elementary knowledge of stereopsis. In this paper, we propose the synthesis method to compose a natural stereo scene with stereoscopic still pictures and moving pictures. Experimental results show that the proposed method in this paper allows a user to synthesize stereoscopic contents easily and compose a stereo scene conveniently.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2000.10a
• /
• pp.176-179
• /
• 2000

The PLL synthesizer is used often in communication system due to several merits, such as broad bandwidth, high accuracy and stability of frequency. But it is difficult to use in current digital communication systems that need frequency hopping at a high speed because of its long frequency hopping time. In this paper, we designed frequency synthesizer that generate the clock frequency randomly at a high speed using the DDS technology and is applied to the pattern generator systemfor for digital image.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2001.05a
• /
• pp.443-447
• /
• 2001

The PLL synthesizer is used often in communication system due to several merits, such as broad bandwidth, high accuracy and stability of frequency. But it is difficult to use in torrent digital communication systems that need frequency hopping at a high speed because of its long frequency hopping time. In this paper, we designed frequency synthesizer that generate the clock frequency randomly at a high speed using the DDS technology and is applied to the pattern generator system for digital image.