• Journal of Broadcast Engineering
• /
• v.13 no.3
• /
• pp.339-349
• /
• 2008

We present a novel method to estimate the light source direction and relight a face texture image of a single 3D model under arbitrary unknown illumination conditions. We create a virtual irradiance sphere to detect the light source direction from a given illuminated texture image using both normal vector mapping and weighted bilinear interpolation. We then induce a relighting equation with estimated ambient and diffuse coefficients. We provide the result of a series of experiments on light source estimation, relighting and face recognition to show the efficiency and accuracy of the proposed method in restoring the shading and shadows areas of a face texture image. Our approach for face relighting can be used for not only illuminant invariant face recognition applications but also reducing visual load and Improving visual performance in tasks using 3D displays.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.30 no.10C
• /
• pp.1036-1043
• /
• 2005

In this paper, we propose a sequential motion vector recovery algorithm for H.264 video coding standard. Motion vectors of H.264 video coding standard cover relatively smaller areas than other standard, since motion estimation of H.264 takes place in the fashion of variable block size. Therefore, the correlation of motion vectors between neighboring blocks increases as the block size of motion estimation is lowered. Under the framework of sequential recovery, we introduce a motion vector recovery using $\alpha$-trimed mean filter. Experimental results show that proposed algorithm is useful in real time video delivery .with nearly comparable or better visual quality than previous approaches such as macro block boundary matching and Lagrage interpolation.

• The KIPS Transactions:PartB
• /
• v.15B no.6
• /
• pp.517-526
• /
• 2008

Stereo vision is a fundamental method for measuring 3D structures by observing them from two cameras placed on different positions. In order to reconstruct 3D structures, it is necessary to create a disparity map from a pair of stereo images. To create a disparity map we compute the matching cost for each point correspondence and compute the disparity that minimizes the sum of the whole matching costs. In this paper, we propose a method to estimate a dense disparity map using region segmentation. We segment each scanline using region homogeneity properties. Using the segmented regions, we prohibit false matches in the stereo matching process. Disparities for pixels that failed in matching are filled by interpolating neighborhood disparities. We applied the proposed method to various stereo images of real environments. Experimental results showed that the proposed method is stable and potentially viable in practical applications.

• Journal of Korea Multimedia Society
• /
• v.16 no.1
• /
• pp.42-49
• /
• 2013

The camera motion is accompanied with the translation and/or the rotation of objects in frames of a video sequence. An unnecessary rotation of objects declines the quality of the moving pictures and in addition is a primary cause of the viewers' fatigue. In this paper, a novel method for correcting rotated frames in video sequences is presented, where the modified Mojette transform is applied to the motion-compensated area in each frame. The Mojette transform is one of discrete Radon transforms, and is modified for correcting the rotated frames as follows. First, the bin values in the Mojette transform are determined by using pixels on the projection line and the interpolation of pixels adjacent to the line. Second, the bin values are calculated only at some area determined by the motion estimation between current and reference frames. Finally, only one bin at each projection is computed for reducing the amount of the calculation in the Mojette transform. Through the simulation carried out on various test video sequences, it is shown that the proposed scheme has good performance for correcting the rotation of frames in moving pictures.

• Journal of IKEEE
• /
• v.25 no.4
• /
• pp.634-643
• /
• 2021

In this paper, an FPGA-based acceleration scheme of range Doppler algorithm (RDA) is proposed for the real time synthetic aperture radar (SAR) imaging. Hardware architectures of matched filter based on systolic array architecture and a high speed sinc interpolator to compensate range cell migration (RCM) are presented. In addition, the proposed hardware was implemented and accelerated on Xilinx Alveo FPGA. Experimental results for 4096×4096-size SAR imaging showed that FPGA-based implementation achieves 2 times acceleration compared to GPU-based design. It was also confirmed the proposed design can be implemented with 60,247 CLB LUTs, 103,728 CLB registers, 20 block RAM tiles and 592 DPSs at the operating frequency of 312 MHz.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.7
• /
• pp.3207-3213
• /
• 2011

Most methods of color constancy, which is the ability to determine the object color regardless of the scene illuminant, have failed to meet our expectation of their performance especially about low-illuminated scenes. Some methods with high performance need to be developed, but we must, above all else, obtain experimental images for analyzing the required circumstances or evaluating the methods. Therefore, the paper produces new sets of images so that they can be used in the development of color constancy methods suitable for low-illuminated scenes. These sets are composed of two parts: one part of images which are synthesized with spectral power distribution(SPD) of illuminants, spectral reflectance curve of reflectances, and sensor response functions of camera; the other part of images where the intensity of each image is adjusted at the uniform rate. In an experiment, the use of the sets takes an advantage that its result images are analyzed and evaluated quantitatively as their ground truth data are known in advance.

• The Journal of Korean Institute of Next Generation Computing
• /
• v.15 no.5
• /
• pp.64-74
• /
• 2019

Although a non-rigid registration has high demands in clinical practice, it has a high computational complexity and it is very difficult for ensuring the accuracy and robustness of registration. This study proposes a method of applying a non-rigid registration to 3D magnetic resonance images of brain in an unsupervised learning environment by using a deep-learning network. A feature vector between two images is produced through the network by receiving both images from two different patients as inputs and it transforms the target image to match the source image by creating a displacement vector field. The network is designed based on a U-Net shape so that feature vectors that consider all global and local differences between two images can be constructed when performing the registration. As a regularization term is added to a loss function, a transformation result similar to that of a real brain movement can be obtained after the application of trilinear interpolation. This method enables a non-rigid registration with a single-pass deformation by only receiving two arbitrary images as inputs through an unsupervised learning. Therefore, it can perform faster than other non-learning-based registration methods that require iterative optimization processes. Our experiment was performed with 3D magnetic resonance images of 50 human brains, and the measurement result of the dice similarity coefficient confirmed an approximately 16% similarity improvement by using our method after the registration. It also showed a similar performance compared with the non-learning-based method, with about 10,000 times speed increase. The proposed method can be used for non-rigid registration of various kinds of medical image data.

• The Journal of the Acoustical Society of Korea
• /
• v.42 no.5
• /
• pp.403-411
• /
• 2023

Functional ultrasound imaging, such as elasticity imaging and micro-blood flow Doppler imaging, enhances diagnostic capability by providing useful mechanical and functional information about tissues. However, the implementation of functional ultrasound imaging poses limitations such as the storage of vast amounts of data in Radio Frequency (RF) data acquisition and processing. In this paper, we propose a sub-Nyquist approach that reduces the amount of acquired axial samples for efficient shear-wave elasticity imaging. The proposed method acquires data at a sampling rate one-third lower than the conventional Nyquist sampling rate and tracks shear-wave signals through RF signals reconstructed using band-pass filtering-based interpolation. In this approach, the RF signal is assumed to have a fractional bandwidth of 67 %. To validate the approach, we reconstruct the shear-wave velocity images using shear-wave tracking data obtained by conventional and proposed approaches, and compare the group velocity, contrast-to-noise ratio, and structural similarity index measurement. We qualitatively and quantitatively demonstrate the potential of sub-Nyquist sampling-based shear-wave elasticity imaging, indicating that our approach could be practically useful in three-dimensional shear-wave elasticity imaging, where a massive amount of ultrasound data is required.

• Journal of Broadcast Engineering
• /
• v.19 no.2
• /
• pp.257-267
• /
• 2014

Distributed compressive video sensing (DCVS) is a low cost sampling paradigm for video coding based on the compressive sensing and the distributed video coding. In this paper, we propose using a skip-mode coding in DCVS under the assumption that in case of high temporal correlation, temporal interpolation can guarantee sufficiently good quality of nonkey frame, therefore no need to transmit measurement data in such a nonkey frame. Furthermore, we extend it to use a hierarchical structure for better temporal interpolation. Simulation results show that the proposed skip-mode coding can save the average subrate of whole video sequence while the PSNR is reduced only slightly. In addition, by using the proposed scheme, the computational complexity is also highly decreased at decoder on average by 43.75% for video sequences that have strong temporal correlation.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.5
• /
• pp.275-282
• /
• 2004

In this paper, a transcoding method that transforms MPEG-4 video bitstream coded in 30 Hz frame rate into H.264 video bitstream of 15 Hz frame rate is proposed. The block modes and motion vectors in MPEG-4 is utilized in H.264 for block mode conversion and motion vector (MV) interpolation methods. The proposed three types of MV interpolation method can be used without performing full motion estimation in H.264. The proposed transcoder reduces computation amount for full motion estimation in H.264 and provides good quality of H.264 video at low bitrates. In experimental results, the proposed methods achieves 3.2-4 times improvement in computational complexity compared to the cascaded pixel-domain transcoding, while the PSNR (peak signal to noise ratio) is degraded with 0.2-0.9dB depending on video sizes.