• Journal of KIISE:Databases
• /
• v.29 no.5
• /
• pp.358-366
• /
• 2002

Feature-based similarity retrieval become an important research issue in image database systems. The features of image data are useful to discrimination of images. In this paper, we propose the high speed k-Nearest Neighbor search algorithm based on Self-Organizing Maps. Self-Organizing Maps(SOM) provides a mapping from high dimensional feature vectors onto a two-dimensional space and generates a topological feature map. A topological feature map preserves the mutual relations (similarities) in feature spaces of input data, and clusters mutually similar feature vectors in a neighboring nodes. Therefore each node of the topological feature map holds a node vector and similar images that is closest to each node vector. We implemented a k-NN search for similar image classification as to (1) access to topological feature map, and (2) apply to pruning strategy of high speed search. We experiment on the performance of our algorithm using color feature vectors extracted from images. Promising results have been obtained in experiments.

• Annual Conference on Human and Language Technology
• /
• 2010.10a
• /
• pp.114-117
• /
• 2010

특정 대상 또는 외부 자극에 대해 반사적이고 직관적으로 발생하는 느낌으로 정의되는 감성은 자신의 경험을 바탕으로 개개인이 서로 다른 반응을 보이는 특징을 가진다. 이러한 감성은 맞춤형, 적응형 및 개인화된 서비스를 요구하는 현대사회에서 반드시 필요하고 연구되어야 하는 대상이다. 하지만, 감성은 외부 자극에 따라 빠르게 변하고 객관성을 유지하기 어려울 뿐 아니라 복합적으로 나타나기 때문에 측정하거나 표현하기가 매우 어렵다. 시각정보는 감성을 이해하고 전달하는 데 큰 비중을 차지하며, 대상에 대한 종합적인 정보를 전달하여 빠르게 인지하고 이해하는데 많은 도움을 준다. 그 중 색채정보는 대상의 객관적인 특정, 심리적 속성 및 사회적 배경을 반영할 수 있어 복합적인 감성을 효과적으로 표현하고 전달한다. 많은 연구를 통해 감성과 색채정보 간의 관계를 생성하고 정의하였지만, 단일 시각정보로 감성을 표현하는 것은 한계가 있다. 본 논문에서는 종합적인 시각정보를 고려한 감성연구를 제안하기 위해 대용량의 감성어휘와 이미지를 수집하였다. 감성어휘는 The Center for the Study of Emotion and Attention(CSEA)에서 생성한 균형 감성어휘 중 273개를 사용하였고 감성이미지는 객관성과 공통성을 유지하기 위해 사용자의 참여가 활발하고 이미지에 부착된 태그가 비교적 정확한 Flickr를 사용하여 수집하였다. 감성어휘 당 약 500개의 이미지를 수집하고자 시도하였고, 총 130,944개의 감성이미지 후보를 수집하였다. 한 번 수집된 이미지는 중복을 피하였고, JPEG형식으로 저장되어 있다. 또한, 각 이미지에는 사용자 태그가 평균적으로 약 25개가 포함되어 있고, 총 2,l47,645개의 태그를 수집하였다.

• Journal of Convergence for Information Technology
• /
• v.11 no.12
• /
• pp.30-38
• /
• 2021

In this paper, we propose a method for extracting thermal data from thermal image and improving detection of heating equipment using the data. The main goal is to read the data in bytes from the thermal image file to extract the thermal data and the real image, and to apply the composite image obtained by synthesizing the image and data to the deep learning model to improve the detection accuracy of the heating facility. Data of KHNP was used for evaluation data, and Faster-RCNN is used as a learning model to compare and evaluate deep learning detection performance according to each data group. The proposed method improved on average by 0.17 compared to the existing method in average precision evaluation.As a result, this study attempted to combine national data-based thermal image data and deep learning detection to improve effective data utilization.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.3
• /
• pp.341-348
• /
• 2020

The ability to extract useful information from an image, such as the human eye, is an interface technology essential for AI computer implementation. The building recognition technology has a lower recognition rate than other image recognition technologies due to the various building shapes, the ambient noise images according to the season, and the distortion by angle and distance. The computer vision based building recognition algorithms presented so far has limitations in discernment and expandability due to manual definition of building characteristics. This paper introduces the deep learning CNN (Convolutional Neural Network) model, and proposes new method to improve the recognition rate even by changes of building images caused by season, illumination, angle and perspective. This paper introduces the partial images that characterize the building, such as windows or wall images, and executes the training with whole building images. Experimental results show that the building recognition rate is improved by about 14% compared to the general CNN model.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.15 no.2
• /
• pp.79-88
• /
• 2022

Social media big data includes a lot of information that can identify not only consumer consumption patterns but also local images. This paper was collected annually data including 'Samcheok' from 2015 to 2019 from Blog and Cafe of Naver and Daum in domestic portal site, and analyzed the regional image change after refining keyword which forms the regional image by performing text mining and network analysis. According to the research results, the regional image of 2015 was expressed with image cognitive elements of the nearby place name or place etc. such as 'Jangho Port', 'Donghae', and 'Beach'. However the regional image both 2016 and 2019 were changed with image cognitive elements of 'SamcheokSolbich' which is a special place within region. Therefore as the keywords related to the local image include 'Jangho Port' and Resort, which are the representative attractions of Samcheok, it can be seen that the infrastructure factor plays a big role in forming the local image. The significance test for the network data used the bootstrap technique, and the p-values in 2015, 2016, and 2019 were 0.0002, 0.0006, and 0.0002, respectively, which were found to be statistically significant at the significance level of 5%.

• Journal of the Korean Applied Science and Technology
• /
• v.40 no.1
• /
• pp.146-159
• /
• 2023

As a makeup technique for a baby-faced image, there will be a difference in perception of the expression technique of baby-faced makeup according to general matters.' Two hypotheses were supported: 'There will be a difference in perception of the expression technique of baby face makeup depending on the general characteristics', and the makeup technique for creating a baby face image is an important function for both men and women, as well as appearance. As a 'physical resource' for social activities, it was confirmed that there is an improvement in the efficiency of the body and mind and an outstanding improvement in mental ability in daily life. Through the results of the study on 'expression of baby face image makeup', awareness and interest in baby face images are high, but research on the production of baby face images is needed. The need for facial expression elements for baby face makeup is expected to be used as basic data for developing baby face images, and this study focuses on external face management for baby face images and baby face makeup.

• Journal of the Korea Computer Graphics Society
• /
• v.29 no.3
• /
• pp.79-84
• /
• 2023

Image generation models have been applied in various fields to overcome data sparsity, time and cost issues. However, it has limitations in generating images from regular pattern images and detecting defects in such data. In this paper, we verified the feasibility of the image generation model to generate pattern images and applied it to data augmentation for defect detection of OLED panels. The data required to train an OLED defect detection model is difficult to obtain due to the high cost of OLED panels. Therefore, even if the data set is obtained, it is necessary to define and classify various defect types. This paper introduces an OLED panel defect data acquisition system that acquires a hypothetical data set and augments the data with an image generation model. In addition, the difficulty of generating pattern images in the diffusion model is identified and a possibility is proposed, and the limitations of data augmentation and defect detection data augmentation using the image generation model are improved.

• Journal of Korean Society of Coastal and Ocean Engineers
• /
• v.35 no.6
• /
• pp.164-172
• /
• 2023

The grain size distribution of sand provides crucial information for understanding coastal erosion and sediment deposition. The commonly used sieve analysis for grain size distribution analysis has limitations such as time-consuming processes and the inability to obtain information about individual particle shapes and colors. In this study, we propose a grain size distribution analysis method using smartphone digital images, which is simpler and more efficient than the sieve analysis method. During the image analysis process, we effectively detect particles from relatively low-resolution smartphone digital images by extracting particle boundaries through image gradient calculation. Using samples collected from four beaches in Gyeongsangbuk-do, we compare and validate the proposed boundary extraction image analysis method with the analysis method that does not extract boundaries, against sieve analysis results. The proposed method shows an average error rate of 8.21% at D₅₀, exhibiting a 65% lower error compared to the method without boundary extraction. Therefore, grain size distribution analysis using smartphone digital images is convenient, efficient, and demonstrated accuracy comparable to sieve analysis.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.2
• /
• pp.13-20
• /
• 2024

This paper introduces a method for generating model images that can identify specific cylindrical medicine containers in videos and investigates data collection techniques. Previous research had separated object detection from specific object recognition, making it challenging to apply automated image stitching. A significant issue was that the coordinate-based object detection method included extraneous information from outside the object area during the image stitching process. To overcome these challenges, this study applies the newly released YOLOv8 (You Only Look Once) segmentation technique to vertically rotating pill bottles video and employs the ORB (Oriented FAST and Rotated BRIEF) feature matching algorithm to automate model image generation. The research findings demonstrate that applying segmentation techniques improves recognition accuracy when identifying specific pill bottles. The model images created with the feature matching algorithm could accurately identify the specific pill bottles.

• Journal of the Korea Society of Computer and Information
• /
• v.29 no.11
• /
• pp.217-227
• /
• 2024

In this paper, we research an image processing method to provide high-capacity image data in real time through a tactical data link and propose a design applicable to the naval combat management system. Image transmission in a tactical data link is important tactically, but real-time performance is limited using existing transmission methods because Image data is larger than other tactical messages. The proposed method improves real-time performance through image processing before and after image transmission, it was designed to improve transmission speed by reducing image data and transmitting the tactical images obtained from naval combat management system, and to ensure that there are no problems with tactical use by restoring the received images with super resolution. The usefulness of this proposed method was confirmed through simulation, and it is expected that it can be applied to various platforms in the future to greatly increase the effectiveness of utilizing tactical images through tactical data link.