• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.39 no.4
• /
• pp.399-407
• /
• 2002

In this paper, a novel algorithm is proposed to reduce the blocking artifacts of block-based coded images by using block classification and MLP. In the proposed algorithm, we classify the block into four classes based on a characteristic of DCT coefficients. And then, according to the class information of neighborhood block, adaptive neural network filter is performed in horizontal and vertical block boundary. That is, for smooth region, horizontal edge region, vertical edge region, and complex region, we use a different two-layer neural network filter to remove blocking artifacts. Experimental results show that the proposed algorithm gives better results than the conventional algorithms both subjectively and objectively.

• Journal of the Korean Society for Precision Engineering
• /
• v.22 no.11 s.176
• /
• pp.135-141
• /
• 2005

Measurement of five DOF motion errors in a ultra precision feed table was attempted in this study. Yaw and pitch error were measured by using a laser interferometer and roll error was measured by using the reversal method. Linear motion errors in the vertical and horizontal directions were measured by using the sequential two point method. In this case, influence of angular motion errors was compensated by using the previously measured ones by the laser interferometer and the reversal method. The capacitive type sensors and an optical straight edge were used in the reversal method and the sequential two point method. Influence of thermal deformation on sensor jig was investgated and minimized by the periodic measurement according to the variation of room temperature. Deviation of gain between sensors was also compensated using the step response data. 5 DOF motion errors of a hydrostatic table driven by the linear motor werer tested using the measurement method. In the horizontal direction, measuring accuracies for the linear and angular motion were within ${\pm}0.02\;{\mu}m$ and ${\pm}0.04$ arcsec, respectively. In the vertical direction, they were within ${\pm}0.02{\mu}m$ and ${\pm}0.05$ arcsec. From these results, it was found that the introduced measurement method was very effective to measure 5 DOF motion errors of the ultra precision feed tables.

• The Journal of the Korea Contents Association
• /
• v.8 no.11
• /
• pp.1-9
• /
• 2008

Thanks to the multimedia technology development, it is possible to show image representations in high quality and to process images in various ways. Anisotropic diffusion as an effective diffusion filtering among many image preprocessing methods and postprocessing methods is used in reduction of speckle noises of ultrasound images, image restoration, edge detection, and image segmentation. However, the conventional anisotropic diffusion based on a cross-kernel causes the following problems. The problem is the concentration of edges in the vertical or horizontal directions. In this paper, a new anisotropic diffusion transform based on directions of gradient is proposed. The proposed method uses the eight directional square-kernel which is an expanded form of the cross-kernel. The proposed method is to select directions of small gradient based on square-kernel. Therefore, the range of proposed diffusion is selected adaptively according to the number of the directions of gradient. Experimental results show that the proposed method can decrease the concentration of edges in the vertical or horizontal directions, remove impulse noise. The image in high quality can be obtained as a result of the proposed method.

• The KIPS Transactions:PartD
• /
• v.16D no.2
• /
• pp.223-236
• /
• 2009

BPMN is a standard business process description notation developed by OMG. It allows the user to have an abstract view of a process that hides its details with the Collapsed Sub-Process notation. While it is a useful direction of abstraction that can be called the horizontal abstraction, a different kind of abstraction, the vertical abstraction, is necessary when different stakeholders of business would like to have different views of the business process form their own viewpoints of interest. For example, stakeholders may want to see a process from the viewpoint of a particular group of actors or from the viewpoint of a certain set of goals. This paper first extends horizontal abstraction capability of BPMN by introducing the notion of super edge and, moreover, adds the vertical abstraction capability to it by introducing the notions of 'aspect attribute' and 'interest specification' and notations for them.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.15 no.6
• /
• pp.709-713
• /
• 2005

A new algorithm is proposed for the extraction of line segments to construct 3D wire-frame models of building from the high-resolution aerial images. The purpose of this study Is the accurate and effective extraction of line segments, considering the problems such as discordance of lines and blurred edges existing in the conventional methods. Using the edge map extracted from aerial images, chain code tracking of edges was performed. Then, we extract the line segments considering the strength of edges and the direction of them. SUSAN (Smallest Uni-value Segment Assimilating Nucleus) algorithm proposed by Smith was used to extract an edge map. The proposed algorithm consists of 4 steps: removal of the horizontal, vertical and diagonal components of edges to reduce non-candidate point of line segments based on the chain code tracking of the edge map, removal of contiguous points, removal of the same angle points, and the extraction of the start and end points to be line segments. By comparing the proposed algorithm with Boldt algorithm, better results were obtained regarding the extraction of the representative line segments of buildings, having relatively less extraction of unnecessary line segments.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.2A
• /
• pp.105-112
• /
• 2012

Face detection at a distance faces is very challenging since images are often degraded by blurring and noise as well as low resolution. This paper proposes an improved face detection method with AdaBoost filtering and sequential testing stages with color and shape information. The conventional AdaBoost filter detects face regions but often generates false alarms. The face detection method is improved by adopting sequential testing stages in order to remove false alarms. The testing stages comprise skin-color test and variable edge-mask filtering. The skin-color filtering is composed of two steps, which involve rectangular window regions and individual pixels to generate binary face clusters. The size of the variable edge-mask is determined by the ellipse which is estimated from the face cluster. The validation of the horizontal and vertical ratio of the mask is also investigated. In the experiments, the efficacy of the proposed algorithm is proved by images captured by a CCTV and a smart-phone

• Journal of IKEEE
• /
• v.7 no.2 s.13
• /
• pp.197-209
• /
• 2003

A current TV out format is quite different from that of HDTV or PC monitor in encoding techniques. In other words, a conventional analog TV uses interlaced display while HDTV or PC monitor uses Non-interlaced / Progressive-scanned display. In order to encode image signals coming from devices that takes interlaced display format for progressive scanned display, a hardware logic in which scanning and interpolation algorithms are implemented is necessary. The ELA(Edge-Based Line Average) algorithm have been widely used because it provided good characteristics. In this study, the ADI(Adaptive De-interlacing Interpolation) algorithm using to improve the ELA algorithm which shows low quality in vertical edge detections and low efficiency of horizontal edge lines. With the De-interlacing ASIC chip that converts the interlaced Digital YUV to De-interlaced Digital RGB is designed. The VHDL is used for chip design.

• Korean Journal of Remote Sensing
• /
• v.22 no.6
• /
• pp.601-612
• /
• 2006

This study proposes RAG (Region Adjancency Graph)-based image segmentation for large imagery in remote sensing. The proposed algorithm uses CN-chain linking for computational efficiency and multi-window operation of sliding structure for memory efficiency. Region-merging due to RAG is a process to find an edge of the best merge and update the graph according to the merge. The CN-chain linking constructs a chain of the closest neighbors and finds the edge for merging two adjacent regions. It makes the computation time increase as much as an exact multiple in the increasement of image size. An RNV (Regional Neighbor Vector) is used to update the RAG according to the change in image configuration due to merging at each step. The analysis of large images requires an enormous amount of computational memory. The proposed sliding multi-window operation with horizontal structure considerably the memory capacity required for the analysis and then make it possible to apply the RAG-based segmentation for very large images. In this study, the proposed algorithm has been extensively evaluated using simulated images and the results have shown its potentiality for the application of remotely-sensed imagery.

• Korean Journal of Remote Sensing
• /
• v.23 no.2
• /
• pp.113-117
• /
• 2007

In this paper, we propose a new method for determining automatically the size of the matching window using histogram of the gradient in order to improve the performance of stereo matching using one-meter resolution satellite imagery. For each pixel, we generate Flatness Index Image by calculating the mean value of the vertical or horizontal intensity gradients of the 4-neighbors of every pixel in the entire image. The edge pixel has high flatness index value, while the non-edge pixel has low flatness index value. By using the histogram of the Flatness Index Image, we find a flatness threshold value to determine whether a pixel is edge pixel or non-edge pixel. If a pixel has higher flatness index value than the flatness threshold value, we classify the pixel into edge pixel, otherwise we classify the pixel into non-edge pixel. If the ratio of the number of non-edge pixels in initial matching window is low, then we consider the pixel to be in homogeneous region and enlarge the size of the matching window We repeat this process until the size of matching window reaches to a maximum size. In the experiment, we used IKONOS satellite stereo imagery and obtained more improved matching results than the matching method using fixed matching window size.

• Journal of KIISE:Software and Applications
• /
• v.37 no.3
• /
• pp.170-176
• /
• 2010

Prediction is an important step in high-performance lossless data compression. In this paper, we propose a novel lossless image coding algorithm to increase prediction accuracy which can display low-resolution images quickly with a multi-resolution image technique. At each resolution, we use pixels of the previous resolution image to estimate current pixel values. For each pixel, we determine its estimated value by considering horizontal, vertical, diagonal edge information and average, weighted-average information obtained from its neighborhood pixels. In the experiment, we show that our method obtains better prediction than JPEG-LS or HINT.