• Proceedings of the Korean Society of Broadcast Engineers Conference
• /
• 2013.06a
• /
• pp.315-318
• /
• 2013

A novel foveation filter method is proposed in DCT domain. For countering the blocking and ringing artifacts, we devise the tools measuring the signal gradient, the block signal variations in the DCT domain. Each measurement is properly applied to each enhancement region and smoothing region. The proposed method optimally adapts the enhancement factors to the characteristics of the underlying signals and so maximizes the enhancement performances with significantly suppressing the artifacts. The subjective and objective evaluations verify that the proposed method sustains producing the improved video qualities for various sequences without tuning any parameters to individual sequences.

• Proceedings of the KSME Conference
• /
• 2001.06b
• /
• pp.700-705
• /
• 2001

The paper presents a seam tracking controller of high speed rotating arc sensor developed by microprocessor based system. The seam tracking algorithm is based on the average current value at each interval region of four phase points on one rotating cycle. To remove the noise effect for the measured current, the area during one rotating cycle is separated into four regions of front, rear, left and right. The average values at each region are calculated, using the regional current values and a low pass filter incorporating the moving average and exponential smoothing methods is adopted.

• Journal of Power System Engineering
• /
• v.6 no.4
• /
• pp.43-48
• /
• 2002

This paper presents a seam tracking controller of high speed rotating arc sensor developed by microprocessor based system. The seam tracking algorithm is based on the average current value at each interval region of four phase points on one rotating cycle. To remove the noise effect for the measured current, the area during one rotating cycle is separated into four regions of front, rear, left and right. The average values at each region are calculated, using the regional current values and a low pass filter incorporating the moving average and exponential smoothing methods is adopted. The effectiveness is proven through the experimental results for several kinds of welding condition.

• Journal of the HCI Society of Korea
• /
• v.2 no.2
• /
• pp.1-9
• /
• 2007

In this paper, we propose a reconstruction technique of the lost hair region for 3D human actor modeling. An active depth sensor system can simultaneously capture both color and geometry information of any objects in real-time. However, it cannot acquire some regions whose surfaces are shiny and dark. Therefore, to get a natural 3D human model, the lost region in depth image should be recovered, especially human hair region. The recovery is performed using both color and depth images. We find out the hair region using color image first. After the boundary of hair region is estimated, the inside of hair region is estimated using an interpolation technique and closing operation. A 3D mesh model is generated after performing a series of operations including adaptive sampling, triangulation, mesh smoothing, and texture mapping. The proposed method can generate recovered 3D mesh stream automatically. The final 3D human model allows the user view interaction or haptic interaction in realistic broadcasting system.

• Journal of the Korea Society of Computer and Information
• /
• v.13 no.6
• /
• pp.243-251
• /
• 2008

This study is to suggest the extraction algorithms of fingerprint corepoint and region separation using the labeling for gate security in order that it might be applied to the fingerprint recognition effectively. The gate security technology is entrance control, attendance management, computer security, electronic commerce authentication, information protection and so on. This study is to extract the directional image by dividing the original image in $128{\times}128$ size into the size of $4{\times}4$ pixel. This study is to separate the region of directional smoothing image extracted by each directional by using the labeling, and extract the block that appeared more than three sorts of change in different directions to the corepoint. This researcher is to increase the recognition rate and matching rate by extracting the corepoint through the separation of region by direction using the maximum direction and labeling, not search the zone of feasibility of corepoint or candidate region of corepoint used in the existing method. According to the result of experimenting with 300 fingerprints, the poincare index method is 94.05%, the proposed method is 97.11%.

• Journal of the Korea Society of Computer and Information
• /
• v.26 no.3
• /
• pp.51-57
• /
• 2021

In this paper, we propose a method to find the DoF(Depth of field) that is blurred in an image by focusing and out-focusing the camera through a efficient convolutional neural network. Our approach uses the RGB channel-based cross-correlation filter to efficiently classify the DoF region from the image and build data for learning in the convolutional neural network. A data pair of the training data is established between the image and the DoF weighted map. Data used for learning uses DoF weight maps extracted by cross-correlation filters, and uses the result of applying the smoothing process to increase the convergence rate in the network learning stage. The DoF weighted image obtained as the test result stably finds the DoF region in the input image. As a result, the proposed method can be used in various places such as NPR(Non-photorealistic rendering) rendering and object detection by using the DoF area as the user's ROI(Region of interest).

• Journal of Biomedical Engineering Research
• /
• v.29 no.2
• /
• pp.122-131
• /
• 2008

For B-mode ultrasound images, we propose an image enhancement algorithm based on a multi-resolution approach, which consists of edge enhancing and noise reducing procedures. Edge enhancement processing is applied sequentially to coarse-to-fine resolution images obtained from wavelet-transformed data. In each resolution, the structural features of each pixel are examined through eigen analysis. Then, if a pixel belongs to an edge region, we perform two-step filtering: that is, directional smoothing is conducted along the tangential direction of the edge to improve continuity and directional sharpening is conducted along the normal direction to enhance the contrast. In addition, speckle noise is alleviated by proper attenuation of the wavelet coefficients of the homogeneous regions at each band. This region-based speckle-reduction scheme is differentiated from other methods that are based on the magnitude statistics of the wavelet coefficients. The proposed algorithm enhances edges regardless of changes in the resolution of an image, and the algorithm efficiently reduces speckle noise without affecting the sharpness of the edge. Hence, compared with existing algorithms, the proposed algorithm considerably improves the subjective image quality without providing any noticeable artifacts.

• Journal of Korea Multimedia Society
• /
• v.21 no.8
• /
• pp.969-981
• /
• 2018

Recent advances in machine learning approaches such as deep neural network and reinforcement learning offer significant performance improvements in generating detailed and varied motions in physically simulated virtual environments. The optimization methods are highly attractive because it allows for less understanding of underlying physics or mechanisms even for high-dimensional subtle control problems. In this paper, we propose an efficient learning method for stochastic policy represented as deep neural networks so that agent can generate various energetic motions adaptively to the changes of tasks and states without losing interactivity and robustness. This strategy could be realized by our novel trajectory search method motivated by the trust region policy optimization method. Our value-based trajectory smoothing technique finds stably learnable trajectories without consulting neural network responses directly. This policy is set as a trust region of the artificial neural network, so that it can learn the desired motion quickly.

• The Journal of the Korea Contents Association
• /
• v.14 no.3
• /
• pp.11-21
• /
• 2014

Multi-scale image segmentation is important in many applications such as image stylization and medical diagnosis. This paper proposes a novel segmentation algorithm based on MSER(maximally stable extremal region) which captures multi-scale structure and is stable and efficient. The algorithm collects MSERs and then partitions the image plane by redrawing MSERs in specific order. To denoise and smooth the region boundaries, hierarchical morphological operations are developed. To illustrate effectiveness of the algorithm's multi-scale structure, effects of various types of LOD control are shown for image stylization. The proposed technique achieves this without time-consuming multi-level Gaussian smoothing. The comparisons of segmentation quality and timing efficiency with mean shift-based Edison system are presented.

• Journal of The Korean Astronomical Society
• /
• v.40 no.4
• /
• pp.133-135
• /
• 2007

A survey project of TRAO with the fifteen beam array receiver system is presented. A multibeam array receiver system has been purchased from FCRAO, and is being installed on TRAO 14m telescope. The target region of the survey is from ${\iota}=120^{\circ}{\sim}137^{\circ},\;b=-1^{\circ}{\sim}+ 1^{\circ}$, and velocity resolution would be 1 km/sec after smoothing from the original resolution of $0.64km\;s^{-1}$ in the transition of J = 1-0 of $^{13}CO$. The survey region is a part of the $^{12}CO$ Outer Galaxy Survey(OGS), and would be an extension of the Bell Laboratories $^{13}CO$ Galactic Plane Survey. By combining with the existing $^{12}CO$ database of the Outer Galaxy Survey, we will derive physical properties of identified molecular clouds and will conduct and statistical analysis of the Outer Galalxy molecular clouds. Reduction process and analysis methods will be introduced.