• The Journal of the Acoustical Society of Korea
• /
• v.17 no.8
• /
• pp.3-11
• /
• 1998

본 논문에서는 고품질의 얼굴 동영상과 운율이 첨가된 음성을 통합하여 자연스런 가상얼굴을 실시간으로 합성하는 방법을 제안한다. 이 방법에서는 한글 텍스트를 입력하여, 텍스트에 따라 입모양과 음성을 합성하고, 얼굴 동영상과 음성의 동기를 맞추고 있다. 먼저, 텍스트를 음운 변화한 후, 문장을 분석하고 자모음사이의 지속시간을 부여한다. 자모음과 지 속시간에 따라 입모양을 변화시켜 얼굴 동영상을 생성하고 있다. 이때, 텍스트에 부합한 입 모양 변화뿐만 아니라, 두부의 3차원 동작과 다양한 표정변화를 통하여 자연스런 가상얼굴 을 실시간으로 합성하고 있다. 한편, 음성합성에서는 문장분석 결과에 따라 강세구와 억양구 를 정하고 있다. 강세구와 억양구를 이용하여 생성된 운율모델이 고품질의 음성합성에 필요 한 지속시간, 억양 및 휴지기를 제어한다. 합성단위는 무제한 어휘가 가능한 반음절과 triphone(VCV)의 조합이며, 합성방식은 TD-PSOLA를 사용한다.

• Journal of Broadcast Engineering
• /
• v.7 no.3
• /
• pp.229-237
• /
• 2002

Natural synthesis of real and virtual images is a key technology in mixed reality. For this purpose, we propose an efficient dense disparity estimation algorithm and a synthesis algorithm considering features of stereo images. Dense disparities are estimated hierarchically from the low to high resolution images. In the process, the region-dividing-bidirectional-matching algorithm makes matching process efficient and keeps the reliability of the estimated disparities, and dense disparities are assigned considering edge information. finally, mixed reality stereo images are synthesized by comparing depth data of real and virtual Images. Computer simulation shows that the proposed algorithms estimate very stable disparity vectors with sharp edge and synthesize natural stereo mixed reality images.

• The KIPS Transactions:PartB
• /
• v.17B no.5
• /
• pp.333-344
• /
• 2010

This paper addresses the way to compose paronamic images from images taken the same objects. With the spread of digital camera, the panoramic image has been studied to generate with its interest. In this paper, we propose a panoramic image generation method using scaling and rotation invariant features. First, feature points are extracted from input images and matched with a RANSAC algorithm. Then, after the perspective model is estimated, the input image is registered with this model. Since the SURF feature extraction algorithm is adapted, the proposed method is robust against geometric distortions such as scaling and rotation. Also, the improvement of computational cost is achieved. In the experiment, the SURF feature in the proposed method is compared with features from Harris corner detector or the SIFT algorithm. The proposed method is tested by generating panoramic images using $640{\times}480$ images. Results show that it takes 0.4 second in average for computation and is more efficient than other schemes.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.1
• /
• pp.67-74
• /
• 2023

The images acquired from the speaker's mouth region revealed the unique patterns according to the corresponding voices. By using this principle, the several methods were proposed in which speech signals were recognized or synthesized from the images acquired at the speaker's lower face. In this study, an image-based speech synthesis method was proposed in which the depth images were cooperatively used. Since depth images yielded depth information that cannot be acquired from optical image, it can be used for the purpose of supplementing flat optical images. In this paper, the usefulness of depth images from the perspective of speech synthesis was evaluated. The validation experiment was carried out on 60 Korean isolated words, it was confirmed that the performance in terms of both subjective and objective evaluation was comparable to the optical image-based method. When the two images were used in combination, performance improvements were observed compared with when each image was used alone.

• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2017.11a
• /
• pp.34-35
• /
• 2017

본 논문에서는 영상으로부터 생성된 깊이맵을 합성곱 신경망(CNN)으로 재생성하는 방법을 제안한다. 합성곱 신경망은 영상인식, 영상분류에 좋은 성능을 보여주는데, 이 기술을 깊이맵 생성에 활용하여 기 제작된 깊이맵 생성 기법을 간단한 합성곱 신경망으로 구현하고자 한다. 성능 실험에서는 10개의 비디오 세트에 제안 방법을 적용한 결과, 만족스러운 결과를 얻었다.

• The KIPS Transactions:PartB
• /
• v.8B no.5
• /
• pp.533-538
• /
• 2001

In this paper, we extend FARGs single mode attribute to multiple attributes for real image application and present a new CBIR using FMARG(Fuzzy Multiple Attribute Relational Graph), which can handle queries involving multiple attributes, not only object label, but also color, texture and spatial relation. In the experiment using the synthetic image database of 1,024 images and the natural image database of 1.026 images built from NETRA database and Corel Draw, the proposed approach shows 6~30% recall increase in the synthetic image database and a good performance, at the displacements and the retrieved number of similar images in the natural image database, compared with the single attribute approach.

• 한국HCI학회:학술대회논문집
• /
• 2007.02a
• /
• pp.494-499
• /
• 2007

실제 영상과 가상의 오브젝트 또는 가상의 환경에 오브젝트를 합성하는 경우 등 사실적인 합성을 결과를 얻기 위해서는 실제 환경과 같은 배경 영상의 정확한 광원 정보가 필요하다. 본 논문에서는 실내 환경을 배경으로 영상을 합성 하는 과정에 필요한 광원정보를 카메라와 Light Probe를 이용하여 촬영된 단일 영상으로부터 추정하는 기법을 제안한다. 실내에 존재하는 광원들은 정확한 위치정보를 알 수 없는 실외환경에서와 달리 제한된 공간의 원점으로부터 3차원 공간에 위치한 좌표로 나타낼 수 있다. 광원을 추정하기 위해 먼저 실내 공간에 반사도가 높은 Light Probe를 위치하고 디지털 카메라의 적정 노출을 이용하여 광원 추정에 사용할 영상을 획득한다. 광원으로 존재하는 오브젝트의 경우 짧은 노출시간에도 카메라의 영상에 획득된다. 그렇기 때문에 단일 영상에서 광원의 영역을 추정하기 위해 영상처리를 통해 노출 시간을 짧게 하여 촬영한 영상과 비슷하게 밝은 영역만 표현되도록 처리를 한다. 전 처리된 영상으로부터 밝은 영역과 어두운 영역으로 구분을 하고 밝은 영역으로부터 광원의 정보를 추정한다. 추정된 광원들은 실제 렌더링에 곧바로 적용이 가능하며, 이를 통해 배경에 적합한 렌더링 결과를 얻을 수 있다.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.1
• /
• pp.193-202
• /
• 2008

In this study, we want to composite the source image and the target image when the environment includes water surface in the target image such as lake, sea, etc. The water surface is different from other common environment. On the water surface, the object must be reflected or refract and sometimes is deformed by the wave of water. In order to composite the object in the source image onto the water image, we analyze the water surface of the target image and let the object be synthesized realistically based on the wave of water. Our composite process consists of three steps. First. we use Shape-from-Shading technique to extract the normal vector of the water surface in the target image. Next, the source image is deformed according to the normal vector map. Finally, we composite the deformed object onto the target image.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.22 no.4
• /
• pp.617-623
• /
• 2018

In this paper, we propose a novel seamless video composition method based on shape matching and Poisson equation. Video composition method consists of video segmentation process and video blending process. In the video segmentation process, the user first sets a trimap for the first frame, and then performs a grab-cut algorithm. Next, considering that the performance of video segmentation may be reduced if the color, brightness and texture of the object and the background are similar, the object region segmented in the current frame is corrected through shape matching between the objects of the current frame and the previous frame. In the video blending process, the object of source video and the background of target video are blended seamlessly using Poisson equation, and the object is located according to the movement path set by the user. Simulation results show that the proposed method has better performance not only in the naturalness of the composite video but also in computational time.

• Progress in Medical Physics
• /
• v.27 no.1
• /
• pp.1-7
• /
• 2016

In this work, we considered a compressed-sensing (CS)-based image deblurring scheme with a total-variation (TV) regularization penalty for improving image characteristics in digital tomosynthesis (DTS). We implemented the proposed image deblurring algorithm and performed a systematic simulation to demonstrate its viability. We also performed an experiment by using a table-top setup which consists of an x-ray tube operated at $90kV_p$, 6 mAs and a CMOS-type flat-panel detector having a $198-{\mu}m$ pixel resolution. In the both simulation and experiment, 51 projection images were taken with a tomographic angle range of ${\theta}=60^{\circ}$ and an angle step of ${\Delta}{\theta}=1.2^{\circ}$ and then deblurred by using the proposed deblurring algorithm before performing the common filtered-backprojection (FBP)-based DTS reconstruction. According to our results, the image sharpness of the recovered x-ray images and the reconstructed DTS images were significantly improved and the cross-plane spatial resolution in DTS was also improved by a factor of about 1.4. Thus the proposed deblurring scheme appears to be effective for the blurring problems in both conventional radiography and DTS and is applicable to improve the present image characteristics.