• Journal of KIISE:Software and Applications
• /
• v.31 no.10
• /
• pp.1386-1399
• /
• 2004

This paper presents correlation-based automatic image captioning. Given a training set of annotated images, we want to discover correlations between visual features and textual features, so that we can automatically generate descriptive textual features for a new unseen image. We develop models with multiple design alternatives such as 1) adaptively clustering visual features, 2) weighting visual features and textual features, and 3) reducing dimensionality for noise sup-Pression. We experiment thoroughly on 10 data sets of various content styles from the Corel image database, about 680MB. The major contributions of this work are: (a) we show that careful weighting visual and textual features, as well as clustering visual features adaptively leads to consistent performance improvements, and (b) our proposed methods achieve a relative improvement of up to 45% on annotation accuracy over the state-of-the-art, EM approach.

• Journal of the Korean Society for information Management
• /
• v.21 no.3
• /
• pp.251-267
• /
• 2004

The purpose of this study is to generate the local level knowledge structure of a single document, similar to end-of-the-book indexes and table of contents of printed material through the use of term clustering and cluster representative term selection. Furthermore, it aims to analyze the functionalities of the knowledge structure. and to confirm the applicability of these methods in user-friend1y information services. The results of the term clustering experiment showed that the performance of the Ward's method was superior to that of the fuzzy K -means clustering method. In the cluster representative term selection experiment, using the highest passage frequency term as the representative yielded the best performance. Finally, the result of user task-based functionality tests illustrate that the automatically generated knowledge structure in this study functions similarly to the local level knowledge structure presented In printed material.

• Journal of Korea Multimedia Society
• /
• v.12 no.2
• /
• pp.272-280
• /
• 2009

The recent mobile industry is providing of a lot of image on/off-line contents are being converted into the mobile contents for architectural design. However, it is difficult to provide users with the existing on/off-line contents without any considerations due to the small size of the mobile screen. In existing methods to overcome the problem, the comic contents on mobile devices are manually produced by computer software such as Photoshop. In this paper, I describe the Automatic Comics Conversion(ACC) system that provides the variedly form of offline comic contents into mobile device of the small screen using Multi-Layer Perceptorn(MLP). ACC produces an experience together with the comic contents fitting for the small screen, which introduces a clustering method that is useful for variety types of comic images and characters as a prerequisite as a stage for preserving semantic meaning. An application is to use the frame form of pictures, website and images in order into mobile device the availability and can bounce back the freeze images contents into dynamic images content.

• The Journal of the Acoustical Society of Korea
• /
• v.16 no.2
• /
• pp.58-65
• /
• 1997

A new automatic classification method with high and stable accuracy for time-series image data is presented in this paper. This method is based on prior condition that a classified map of the target area already exists, or at least one of the time-series image data had been classified. The classified map is used as a reference map to specify training areas of classification categories. The new automatic classification method consists of five steps, i.e., extraction of training data using reference map, detection of changed pixels based upon the homogeneity of training data, clustering of changed pixels, reconstruction of training data, and classification as like maximum likelihood classifier. In order to evaluate the performance of this method qualitatively, four time-series Landsat TM image data were classified by using this method and a conventional method which needs a skilled operator. As a results, we could get classified maps with high reliability and fast throughput, without a skilled operator.

• Journal of Broadcast Engineering
• /
• v.21 no.3
• /
• pp.425-435
• /
• 2016

In this paper, we propose a classification model by analyzing raw material images recorded using a color CCD camera to automatically classify good and defective agricultural products such as rice, coffee, and green tea, and raw materials. The current classifying agricultural products mainly depends on visual selection by skilled laborers. However, classification ability may drop owing to repeated labor for a long period of time. To resolve the problems of existing human dependant commercial products, we propose a vision based automatic raw material classification combining mean shift clustering and stepwise region merging algorithm. In this paper, the image is divided into N cluster regions by applying the mean-shift clustering algorithm to the foreground map image. Second, the representative regions among the N cluster regions are selected and stepwise region-merging method is applied to integrate similar cluster regions by comparing both color and positional proximity to neighboring regions. The merged raw material objects thereby are expressed in a 2D color distribution of RG, GB, and BR. Third, a threshold is used to detect good and defective products based on color distribution ellipse for merged material objects. From the results of carrying out an experiment with diverse raw material images using the proposed method, less artificial manipulation by the user is required compared to existing clustering and commercial methods, and classification accuracy on raw materials is improved.

• Journal of Navigation and Port Research
• /
• v.47 no.4
• /
• pp.231-238
• /
• 2023

The position of a light buoy is always flexible due to the influence of external forces such as tides and wind. The position can be checked through AIS (Automatic Identification System) or RTU (Remote Terminal Unit) for AtoN. As a result of analyzing the position data for the last five years (2017-2021) of a light buoy, the average position error was 15.4%. It is necessary to detect position error data and obtain refined position data to prevent navigation safety accidents and management. This study aimed to detect position error data and obtain refined position data by DBSCAN Clustering position data obtained through AIS or RTU for AtoN. For this purpose, 21 position data of Gunsan Port No. 1 light buoy where RTU was installed among western waters with the most position errors were DBSCAN clustered using Python library. The minPts required for DBSCAN Clustering applied the value commonly used for two-dimensional data. Epsilon was calculated and its value was applied using the k-NN (nearest neighbor) algorithm. As a result of DBSCAN Clustering, position error data that did not satisfy minPts and epsilon were detected and refined position data were acquired. This study can be used as asic data for obtaining reliable position data of a light buoy installed with AIS or RTU for AtoN. It is expected to be of great help in preventing navigation safety accidents.

• Journal of the Korean Institute of Intelligent Systems
• /
• v.12 no.6
• /
• pp.577-582
• /
• 2002

Neural network can be trained to approximate an arbitrary nonlinear function of multivariate data like the mini-mill crown values in Automatic Shape Control. The trained weights of neural network can evaluate or generalize the process data outside the training vectors. Sometimes, the blind modeling of the process data is necessary to compare with the scattered analytical model of mini-mill process in isolated electro-mechanical forms. To come up with a viable model, we propose the blind neural-based range-division domain-clustering piecewise-linear modeling scheme. The basic ideas are: 1) dividing the range of target data, 2) clustering the corresponding input space vectors, 3)training the neural network with clustered prototypes to smooth out the convergence and 4) solving the resulting matrix equations with a pseudo-inverse to alleviate the ill-conditioning problem. The simulation results support the effectiveness of the proposed scheme and it opens a new way to the data analysis technique. By the comparison with the statistical regression, it is evident that the proposed scheme obtains better modeling error uniformity and reduces the magnitudes of errors considerably. Approximatly 10-fold better performance results.

• Annual Conference on Human and Language Technology
• /
• 1998.10c
• /
• pp.419-424
• /
• 1998

본 논문에서는 단어들의 분포적 특성을 이용하여 자동으로 단어를 군집화(clustering) 하는 기법을 제시한다. 제안된 군집화 기법에서는 단어들 사이의 거리(distance)를 가상 공간상에 있는 두 단어의 평균점에 대한 불일치의 합(total divergence to the average)으로 측정하며 군집화 알고리즘으로는 최소 신장 트리(minimal spanning tree)를 이용한다. 본 논문에서는 이 기법에 대해 두 가지 실험을 수행한다. 첫 번째 실험은 코퍼스에서 상위 출현 빈도를 가지는 약 1,200 개의 명사들을 의미에 따라 군집화 하는 것이며 두 번째 실험은 이 논문에서 제시한 자동 군집화 방법의 성능을 객관적으로 평가하기 위한 것으로 가상 단어(pseudo word)에 대한 군집화이다. 실험 결과 이 방법은 가상 단어에 대해 약 91%의 군집화 정확도와(clustering precision)와 약 81%의 군집 순수도(cluster purity)를 나타내었다. 한편 두 번째 실험에서는 평균점에 대한 불일치의 합을 이용한 거리 측정에서 나타나는 문제점을 보완한 거리 측정 방법을 제시하였으며 이를 이용하여 가상 단어 군집화를 수행한 결과 군집화 정확도와 군집 순수도가 각각 약 96% 및 95%로 향상되었다.

• Journal of Korea Multimedia Society
• /
• v.11 no.8
• /
• pp.1169-1178
• /
• 2008

Most of information in Web2.0 is constructed by users and can be classified by tags which are also constructed and added by users. However, as we known, referring by the related works such as automatic tagging techniques and tag cloud's construction techniques, the research to be classified information and resources by tags effectively is to be given users which is still up to the mark. In this paper, we propose and implement a clustering system that does mapping each other according to relationships of the resource's tags collected from Web and then makes the mapping result into clusters to retrieve images. Tn addition, we analyze our system's efficiency by comparing our proposed system's image retrieval result with the image retrieval results searched by Flickr website.

• Journal of the Korean Society for Precision Engineering
• /
• v.16 no.5 s.98
• /
• pp.48-57
• /
• 1999

Applications of thresholding technique are based on the assumption that object and background pixels in a digital image can be distinguished by their gray level values. For the segmentation of more complex images, it is necessary to resort to multiple threshold selection techniques. This paper describes a new method for multiple threshold selection of gray level images which are not clearly distinguishable from the background. The proposed method consists of three main stages. In the first stage, a probability distribution function for a gray level histogram of an image is derived. Cluster points are defined according to the probability distribution function. In the second stage, fuzzy partition matrix of the probability distribution function is generated through the fuzzy clustering process. Finally, elements of the fuzzy partition matrix are classified as clusters according to gray level values by using max-membership method. Boundary values of classified clusters are selected as multiple threshold. In order to verify the performance of the developed algorithm, automatic inspection process of ball grid array is presented.