• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.24 no.5
• /
• pp.674-677
• /
• 2020

The cultural background of users utilizing image search engines has a significant impact on the satisfaction of the search results. Therefore, it is important to analyze and understand the cultural context of images for more accurate image search. In this paper, we investigate how the cultural context of images can affect the performance of image classification. To this end, we first collected various types of images (e.g,. food, temple, etc.) with various cultural contexts (e.g., Korea, Japan, etc.) from web search engines. Afterwards, a deep transfer learning approach using VGG19 and MobileNetV2 pre-trained with ImageNet was adopted to learn the cultural features of the collected images. Through various experiments we show the performance of image classification can be differently affected according to the cultural context of images.

• Journal of the Semiconductor & Display Technology
• /
• v.21 no.4
• /
• pp.156-162
• /
• 2022

Image compression plays an important role in encoding and improving various forms of images in the digital era. Recent researches have focused on the principle of deep learning as one of the most exciting machine learning methods to show that it is good scheme to analyze, classify and compress images. Various neural networks are able to adapt for image compressions, such as deep neural networks, artificial neural networks, recurrent neural networks and convolution neural networks. In this review paper, we discussed how to apply the rule of deep learning to obtain better image compression with high accuracy, low loss-ness and high visibility of the image. For those results in performance, deep learning methods are required on justified manner with distinct analysis.

• Proceedings of the KSRS Conference
• /
• 2005.10a
• /
• pp.138-141
• /
• 2005

In this paper, we propose a watermarking technique of multispectral images. In our method, the Principal Component Analysis (PCA) is preliminarily applied on the multispectral image. The most principal component image is used for embedding with a watermark, which is a pseudo-random number sequence generated with a secret key. The embedding process is performed in the wavelet domain. The resulting image is then reinserted into the principal component images, and the final multispectral image containing the watermark can be produced by the inverse PCA. Experimental results are provided to illustrate the performance of the algorithm against various kinds of attacks.

• Journal of Mechanical Science and Technology
• /
• v.18 no.6
• /
• pp.879-894
• /
• 2004

The image processing technique is simple and, in principle, can handle particles with various shapes since it is based on direct visualization. Moreover, a wide measurement area can be covered with appropriate optical arrangement. In the present paper, various techniques of image processing for sizing and counting particles are reviewed and recent developments are introduced. Two major subjects are discussed in detail: identification of particles (i.e., boundary detection and pattern recognition) and determination of in-focus criteria. Finally, an overall procedure for image processing of spray particles is suggested.

• 제어로봇시스템학회:학술대회논문집
• /
• 2005.06a
• /
• pp.2459-2463
• /
• 2005

Optical coherence tomography (OCT) is high resolution medical imaging system which is obtaining image inside biological objects with non-destructive method. OCT system is based on Michelson interferometer with a reciprocating mirror in the reference arm and a biological object in the sample arm. The obtained OCT image suffers from a granular or mottled image, called speckle. Speckle is caused by random interferences between reflected coherence waves. In this paper, we propose effective speckle reduction method that uses wavelet transform. With wavelet domain image, sub-windowing and thresholding are performed. Finally, speckle reduction experiments for Misgurnus mizolepis skin and rat eye images are shown.

• ETRI Journal
• /
• v.35 no.2
• /
• pp.207-217
• /
• 2013

This paper introduces the modular cellular neural network (CNN), which is a new CNN structure constructed from nine one-layer modules with intercellular interactions between different modules. The new network is suitable for implementing many image processing operations. Inputting an image into the modules results in nine outputs. The topographic characteristic of the cell interactions allows the outputs to introduce new properties for image processing tasks. The stability of the system is proven and the performance is evaluated in several image processing applications. Experiment results on texture segmentation show the power of the proposed structure. The performance of the structure in a real edge detection application using the Berkeley dataset BSDS300 is also evaluated.

• Journal of the Semiconductor & Display Technology
• /
• v.7 no.2
• /
• pp.69-76
• /
• 2008

Recently, as the multimedia processing application increases rapidly by going on increasing multimedia data, the efficient retrieval method of image information is required in many fields of application and becoming the matter of major concern. Furthermore, in the last few years rapid improvements in hardware technology have made it possible to process, store and retrieve huge amounts of data in a multimedia format. As a result, Content-Based Image Retrieval (CBIR) has been receiving widespread interest during the last decade. This paper propose the content-based retrieval system as a method for performing image retrieval through the effective feature analysis of the object of significant meaning by using YCbCr channel merging on the basis of the characteristics of man's visual system.

• Proceedings of the KIEE Conference
• /
• summer
• /
• pp.1409-1411
• /
• 2004

[$80\;\sim\;90\%$ of Information that human masters from outside are image information which depend on sight. So, in case machine takes the place of human's function, role of image processing is very important. However, because human's sight function is formed through evolution and studying for a long time, substituting it for computer attend with unexpected difficulty. In railroad, there are many field which the application of automation system is expected by using computer instead of safety watch that depend on usual human sight, macrography. But, there are many research field and conditions in the railroad when various outside conditions are considered, comparing with another field. This study analyses the recent image processing technology and reviews the task of image processing technology and its direction.]

• Proceedings of the KSRS Conference
• /
• 2003.11a
• /
• pp.222-224
• /
• 2003

Many users want to add some their own information to data which was on the web and computer without actually needing to touch data. In remote sensing, the result data for image classification consist of image and text file in general. To overcome these inconvenience problems, we suggest the annotation method using XML language. We give the efficient annotation method which can be applied to web and viewing of image classification. We can apply the annotation for web and image classification with image and text file. The need for thesaurus construction is the lack of information for remote sensing and GIS on search engine like Empas, Naver and Google. In search engine, we can’t search the information for word which has many different names simultaneously. We select the remote sensing data from different sources and make the relation between many terms. For this process, we analyze the meaning for different terms which has similar meaning.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.16 no.10
• /
• pp.3406-3418
• /
• 2022

Reversible data hiding is an algorithm that has ability to extract the secret data and to restore the marked image to its original version after data extracting. However, some previous schemes offered the low image quality of marked images. To solve this shortcoming, a new reversible data hiding scheme based on pixel value ordering and edge detection mechanism is proposed. In our proposed scheme, the edge image is constructed to divide all pixels into the smooth regions and rough regions. Then, the pixels in the smooth regions are separated into non overlapping blocks. Then, by taking advantages of the high correlation of current pixels and their adjacent pixels in the smooth regions, PVO algorithm is applied for embedding secret data to maintain the minimum distortion. The experimental results showed that our proposed scheme obtained the larger embedding capacity. Moreover, the greater image quality of marked images are achieved by the proposed scheme than that other previous schemes while the high EC is embedded.