• Journal of Information Processing Systems
• /
• v.7 no.3
• /
• pp.447-458
• /
• 2011

This study presents a comprehensive solution to the collection management, which is based on the model for Cultural Objects (CCO). The developed system manages and spreads the collections that are safeguarded in museums and galleries more easily by using IT. In particular, we present our approach for a non-structured search and recovery of the objects based on the annotation of artwork images. In this methodology, we have introduced a faceted search used as a framework for multi-classification and for exploring/browsing complex information bases in a guided, yet unconstrained way, through a visual interface.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.1990-1993
• /
• 2005

Stereovision is known to be one of the most important tools for robot vision systems. Previously, 2D active contour model has been applied to stereovision by defining the contour on the 3D space instead of image plane. However, the proposed model is still that of curve so that some complex shapes such as surfaces with high curvature can not be properly estimated because of occlusion phenomena. In this paper, the authors extend the curve model to the surface model. The surface is approximated by polygons and new energy function and its optimization method for surface estimation is proposed. Its effectiveness is examined by experiments with real stereo images.

• Journal of the Korean Institute of Telematics and Electronics B
• /
• v.31B no.5
• /
• pp.141-148
• /
• 1994

Both the neural network classifier utilizing multi-layer perceptron and the linear tree classifier composed of hierarchically structured linear discriminating functions can form arbitrarily complex decision boundaries in the feature space and have very similar decision making processes. In this paper, a new method for automatically choosing the number of neurons in the hidden layers and for initalzing the connection weights between the layres and its supporting theory are presented by mapping the sequential structure of the linear tree classifier to the parallel structure of the neural networks having one or two hidden layers. Experimental results on the real data obtained from the military ship images show that this method is effective, and that three exists no siginificant difference in the classification acuracy of both classifiers.

• 한국정보디스플레이학회:학술대회논문집
• /
• 2008.10a
• /
• pp.745-748
• /
• 2008

The theory of Bartleson and Breneman that the perceived image contrast changes with surround luminance (the lighter surround provides higher contrast) was tested an over bright condition($8500d/m^2$). Contrarily to the Bartleson and Breneman's results, we observed the fact that perceived constrast was decreased when surround huminance increased from dark to over bright through two sets of psychophysical experiments based upon both uniform gray patches and complex color images.

• The Transactions of the Korea Information Processing Society
• /
• v.7 no.8
• /
• pp.2575-2586
• /
• 2000

In this paper, we propose an algorithm which can automatically extract the road information in a map image. The proposed method extracts the road image in the complex map image. The extracted image is converted into the skeleton image by thining method. The converted image contains various problems. In order to correct these problems, after the road is classified by the number of Rutovitz-connectivity which represents the characteristic of road, those are respectively corrected according to the load characteristic. In the simulation, the proposed method has obtained good results for the various type of map images.

• IEIE Transactions on Smart Processing and Computing
• /
• v.2 no.1
• /
• pp.1-10
• /
• 2013

In JPEG2000, the Cohen-Daubechies-Feauveau (CDF) 9/7-tap wavelet filter is implemented using the conventional lifting scheme. On the other hand, this wavelet filter has two problems: the filter coefficients remain complex, and the conventional lifting scheme does not consider the image edges in the coding process. This paper proposes an effective lifting scheme to solve these problems. For this purpose, optimal 9/7-tap wavelet filters were designed in two steps. In the first step, the appropriate filter coefficients were selected. In the second step, a median operator was employed to consider the image edges. The experimental results with the median lifting scheme and the combination of filter optimization with the median lifting show that the proposed methods outperform the well-known CDF 9/7-tap wavelet filter of JPEG2000 on the edge-dominant images.

• International Journal of Control, Automation, and Systems
• /
• v.1 no.2
• /
• pp.229-235
• /
• 2003

We designed a family of completely autonomous mobile robots with local intelligence. Each robot has a number of on-board sensors, including vision, and does not rely on global positioning systems The on-board embedded controller is sufficient to analyze several low-resolution color images per second. This enables our robots to perform several complex tasks such as navigation, map generation, or providing intelligent group behavior. Not being limited to playing the game of soccer and being completely autonomous, we are also looking at a number of other interesting scenarios. The robots can communicate with each other, e.g. for exchanging positions, information about objects or just the local states they are currently in (e.g. sharing their current objectives with other robots in the group). We are particularly interested in the differences between a behavior-based approach versus a traditional control algorithm at this still very low level of action.

• Proceedings of the Korea Electromagnetic Engineering Society Conference
• /
• 2000.11a
• /
• pp.117-120
• /
• 2000

When analyzing the scatting problem of multi-layered structures using closed-form Green's function, one of the main difficulties is that the numerical integrations for the evaluation of diagonal matrix elements converge slowly and are not so stable. Accordingly, even when the integration for the singularity of type e$\^$-jkr//${\gamma}$/, corresponding to the source dipole itself, is performed using such a mathod, this difficulty persists in the integration corresponding to the finite number of complex images. In order to resolve this difficulty, a new technique based upon the Gaussian quadrature in polar coordinates for the evaluation of the two-dimensional generalized exponential integral is presented. Stability of the algorithm and convergence is discussed. Performance is demonstrated for the example of a microstrip patch antenna.

• The Bulletin of The Korean Astronomical Society
• /
• v.42 no.2
• /
• pp.41.1-41.1
• /
• 2017

We present weak-lensing analysis of the merging galaxy cluster Abell 115 at z=0.197 based on Subaru i and V band images. As merging clusters often show, Abell 115's merging signatures include radio relics, double X-ray peaks, and large offsets between the cluster member galaxies and the X-ray distributions. A weak-lensing study provides underlying dark matter distribution, the key information to determine the complex merging scenario of the cluster. In this work, we present 2D mass reconstruction of the cluster, which reveals two distinct mass peaks consistent with galaxy distributions. We measure the first weak-lensing mass of each subcluster. Our weak-lensing total mass estimate is a few factors lower than the published dynamical mass obtained from velocity dispersion. This large mass discrepancy may be attributed to a significant departure from dynamical equilibrium. We also re-analyze the archival chandra data and find that the result is consistent with weak-lensing mass.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.8
• /
• pp.3534-3549
• /
• 2020

The traditional method of smoke image recognition has low accuracy. For this reason, we proposed an algorithm based on the good group of IMFOA which is GMFOA to optimize the parameters of SVM. Firstly, we divide the motion region by combining the three-frame difference algorithm and the ViBe algorithm. Then, we divide it into several parts and extract the histogram of oriented gradient and volume local binary patterns of each part. Finally, we use the GMFOA to optimize the parameters of SVM and multiple kernel learning algorithms to Classify smoke images. The experimental results show that the classification ability of our method is better than other methods, and it can better adapt to the complex environmental conditions.