• The Bulletin of The Korean Astronomical Society
• /
• v.38 no.2
• /
• pp.98-98
• /
• 2013

태양의 활동영역에서 관측할 수 있는 흑점은 주로 흑점군으로 관측되며, 태양폭발현상의 발생을 예보하기 위한 중요한 관측 대상 중 하나이다. 현재 태양 폭발을 예보하는 모델들은 McIntosh 흑점군 분류법을 사용하며 통계적 모델과 기계학습 모델로 나누어진다. 컴퓨터는 흑점군의 형태학적 특성을 연속적인 값으로 계산하지만 흑점군의 형태적 다양성으로 인해 McIntosh 분류법과 일치하지 않는 경우가 있다. 이러한 이유로 컴퓨터가 계산한 흑점군의 형태학적인 특성을 예보에 직접 적용하는 것이 필요하다. 우리는 흑점군을 검출하기 위해 최소신장트리(Minimum spanning tree : MST)를 이용한 계층적 군집화 기법을 수행하였다. 그래프(Graph)이론에서 최소신장트리는 정점(Vertex)과 간선(Edge)으로 구성된 간선의 가중치의 합이 최소인 트리이다. 우리는 모든 흑점을 정점, 그들의 연결을 간선으로 적용하여 최소신장트리를 작성하였다. 또한 최소신장트리를 활용한 계층적 군집화기법은 초기값에 따른 군집화 결과의 차이가 없기 때문에 흑점군 검출에 있어서 가장 적합한 알고리즘이다. 이를 통해 흑점군의 기본적인 형태학적인 특성(개수, 면적, 면적비 등)을 계산하고 최소신장트리를 통해 가장 면적이 큰 흑점을 중심으로 트리의 깊이(Depth)와 차수(Degree)를 계산하였다. 이 방법을 2003년 SOHO/MDI의 태양 가시광 영상에 적용하여 구한 흑점군의 내부 흑점수와 면적은 NOAA에서 산출한 값들과 각각 90%, 99%의 좋은 상관관계를 가졌다. 우리는 이 연구를 통해 흑점군의 형태학적인 특성과 더불어 예보에 직접적으로 활용할 수 있는 방법을 논의하고자 한다.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.26 no.10
• /
• pp.1441-1446
• /
• 2022

Modern people can access various information about food recipes very easily on the Internet or social media. As the supply of food recipes increases, it is difficult to find a suitable recipe for each user in the overflowing information. As such, the need to provide information by reflecting users' requirements has increased, and research related to food recipes and cooking recommendations is becoming active. In addition, the Internet, video, and application markets using this are also rapidly activating. In this study, in order to classify recipes from the user's perspective of food recipe users, the user's review data was applied with the k-mean clustering technique, which is unsupervised learning, and a "food recipe classification model" was derived. As a result, it was classified into a total of 25 clusters including information needed by many users, such as specific purposes and cooking stages.

• Journal of the Korea Society of Computer and Information
• /
• v.16 no.1
• /
• pp.61-69
• /
• 2011

This paper proposes a new algorithm that improve color component of compensated image using Retinex method for back-light image. A back-light image has two regions, one of the region is too bright and the other one is too dark. If an back-light image is improved contrast using Retinex method, it loses color information in the part of brightness of the image. In order to make up loss information, proposed algorithm adds color components from original image. The histogram can be divided three parts that brightness, darkness, midway using K-mean (k=3) algorithm. For the brightness, it is used color information of the original image. For the darkness, it is converted using by Retinex method. The midway region is mixed between original image and Retinex result image in the ratio of histogram. The ratio is determined by distance from dark area. The proposed algorithm was tested on nature back-light images to evaluate performance, and the experimental result shows that proposed algorithm is more robust than original Retinex algorithm.

• The KIPS Transactions:PartB
• /
• v.12B no.4 s.100
• /
• pp.369-378
• /
• 2005

Document image is segmented and classified into text, picture, or table by a document layout analysis, and the words in table regions are significant for keyword spotting because they are more meaningful than the words in other regions. This paper proposes a method to extract words from table regions in document images. As word extraction from table regions is practically regarded extracting words from cell regions composing the table, it is necessary to extract the cell correctly. In the cell extraction module, table frame is extracted first by analyzing connected components, and then the intersection points are extracted from the table frame. We modify the false intersections using the correlation between the neighboring intersections, and extract the cells using the information of intersections. Text regions in the individual cells are located by using the connected components information that was obtained during the cell extraction module, and they are segmented into text lines by using projection profiles. Finally we divide the segmented lines into words using gap clustering and special symbol detection. The experiment performed on In table images that are extracted from Korean documents, and shows $99.16\%$ accuracy of word extraction.

• Journal of the Korea Society of Computer and Information
• /
• v.15 no.1
• /
• pp.91-102
• /
• 2010

It is very difficult that a previous surveillance system and vehicle detection system find objects on a limited and unstable illumination condition. This paper proposes a robust method of adaptively updating a reference background image for solving problems that are generated by the unstable illumination. The first input image is set up as the reference background image, and is divided into three block categories according to an edge component. Then a block state analysis, which uses a rate of change of the brightness, a stability, a color information, and an edge component on each block, is applied to the input image. On the reference background image, neighbourhood blocks having the same state of a updated block are merged as a block. The proposed method can generate a robust reference background image because it distinguishes a moving object area from an unstable illumination. The proposed method very efficiently updates the reference background image from the point of view of the management and the processing time. In order to demonstrate the superiority of the proposed stable manner in situation that an illumination quickly changes.

• Journal of the Korea Computer Graphics Society
• /
• v.25 no.3
• /
• pp.123-131
• /
• 2019

In this study, we modified CT images of femoral head in consideration of anatomically meaningful structure, proposing the method to augment the training data of convolution Neural network for segmentation of femur mesh model. First, the femur mesh model is obtained from the CT image. Then divide the mesh model into meaningful parts by using cluster analysis on geometric characteristic of mesh surface. Finally, transform the segments by using an appropriate mesh deformation algorithm, then create new CT images by warping CT images accordingly. Deep learning models using the data enhancement methods of this study show better image division performance compared to data augmentation methods which have been commonly used, such as geometric conversion or color conversion.

• Journal of Broadcast Engineering
• /
• v.22 no.5
• /
• pp.643-651
• /
• 2017

Steganography is one of information hiding technologies and discriminated from cryptography in that it focuses on avoiding the existence the hidden information from being detected by third parties, rather than protecting it from being decoded. In this paper, as an image steganography method which uses images as media, we propose a new block matching method that embeds information into the discrete wavelet transform (DWT) domain. The proposed method, based on a statistical analysis, reduces loss of embedding capacity due to inequable use of candidate blocks. It works in such a way that computes the variance of each candidate block, preserves candidate blocks with high frequency components while reducing candidate blocks with low frequency components by compressing them exploiting the k-means clustering algorithm. Compared with the previous block matching method, the proposed method can reconstruct secret images with similar PSNRs while embedding higher-capacity information.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.1
• /
• pp.139-148
• /
• 2015

In this paper, a new algorithm is proposed for detecting the product number of each thick plate and extracting each character of the product number from a image which contains several thick plates. In general, a image of thick plates contains several steal plates. To obtain the product number from the image, we first need to separate each plate. To do so, we use the line edges of thick plates and a clustering algorithm. After separating each plate, background parts are eliminated from the image of each plate. Background parts of an individual thick plate image consist of the dark part of steel and the white part of paint which is used for printing the product number. We propose a two-tiered method where dark background parts are first eliminated and then white parts are eliminated. Finally, each character is extracted from the product number image using the characteristics of product number. The results of the experiments on the various steal plates images emphasize that the proposed algorithm detects each thick plate and extracts the product number from a image effectively.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.47 no.5
• /
• pp.192-198
• /
• 2010

With the popularization of the mobile phone with a built-in camera, there are a lot of effort to provide useful information to users by detecting and recognizing the text in the video which is captured by the camera in mobile phone, and there is a need to detect the text regions in such mobile phone video. In this paper, we propose a method to detect the text regions in the mobile phone video. We employ morphological operation as a preprocessing and obtain binarized image using modified k-means clustering. After that, candidate text regions are obtained by applying connected component analysis and general text characteristic analysis. In addition, we increase the precision of the text detection by examining the frequency of the candidate regions. Experimental results show that the proposed method detects the text regions in the mobile phone video with high precision and recall.

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.34 no.4C
• /
• pp.388-396
• /
• 2009

This paper proposes a new fire-smoke detection method by using extracted features from camera images and pattern recognition technique. First, moving regions are detected by analyzing the frame difference between two consecutive images and generate candidate smoke regions by applying smoke color model. A smoke region generally has a few characteristics such as similar color, simple texture and upward motion. From these characteristics, we extract brightness, wavelet high frequency and motion vector as features. Also probability density functions of three features are generated using training data. Probabilistic models of smoke region are then applied to observation nodes of our proposed Dynamic Bayesian Networks (DBN) for considering time continuity. The proposed algorithm was successfully applied to various fire-smoke tasks not only forest smokes but also real-world smokes and showed better detection performance than previous method.