• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.49 no.2
• /
• pp.94-101
• /
• 2012

This paper proposes an improved image steganalysis scheme to raise the detection rate of stego images out of cover images. To improve the detection rate of stego image in the steganalysis, tiny variation caused by data hiding should be amplified. For this, we extract feature vectors of cover image and stego image by two steps. First, we separate image into upper 4 bit subimage and lower 4 bit subimage. As a result, stego noise is expanded more than two times. We decompose separated subimages into twelve subbands by applying 3-level Haar wavelet transform and calculate co-occurrence probabilities of two different subbands in the same scale. Since co-occurrence probability of the two wavelet subbands is affected by data hiding, it can be used as a feature to differentiate cover images and stego images. The extracted feature vectors are used as the input to the multilayer perceptron(MLP) classifier to distinguish between cover and stego images. We test the performance of the proposed scheme over various embedding rates by the LSB, S-tool, COX's SS, and F5 embedding method. The proposed scheme outperforms the previous schemes in detection rate to existence of hidden message as well as exactness of discrimination.

• The KIPS Transactions:PartD
• /
• v.15D no.6
• /
• pp.755-766
• /
• 2008

Recently, NAND flash memory is becoming into the spotlight as a next-generation storage device because of its small size, fast speed, low power consumption, and etc. compared to the hard disk. However, due to the distinct characteristics such as erase-before-write architecture, asymmetric operation speed and unit, disk-based systems and applications may result in severe performance degradation when directly implementing them on NAND flash memory. Especially when a B-tree is implemented on NAND flash memory, intensive overwrite operations may be caused by record inserting, deleting, and reorganizing. These may result in severe performance degradation. Although ${\mu}$-tree has been proposed in order to overcome this problem, it suffers from frequent node split and rapid increment of its height. In this paper, we propose Log-Structured B-Tree(LSB-Tree) where the corresponding log node to a leaf node is allocated for update operation and then the modified data in the log node is stored at only one write operation. LSB-tree reduces additional write operations by deferring the change of parent nodes. Also, it reduces the write operation by switching a log node to a new leaf node when inserting the data sequentially by the key order. Finally, we show that LSB-tree yields a better performance on NAND flash memory by comparing it to ${\mu}$-tree through various experiments.

• KIPS Transactions on Software and Data Engineering
• /
• v.2 no.8
• /
• pp.579-588
• /
• 2013

This paper proposes an effective watermarking approach for copyright protection of fire video images. The proposed watermarking approach efficiently utilizes the inherent characteristics of fire data with respect to color and texture by using a gray level co-occurrence matrix (GLCM) and fuzzy c-means (FCM) clustering. GLCM is used to generate a texture feature dataset by computing energy and homogeneity properties for each candidate fire image block. FCM is used to segment color of the fire image and to select fire texture blocks for embedding watermarks. Each selected block is then decomposed into a one-level wavelet structure with four subbands [LL, LH, HL, HH] using a discrete wavelet transform (DWT), and LH subband coefficients with a gain factor are selected for embedding watermark, where the visibility of the image does not affect. Experimental results show that the proposed watermarking approach achieves about 48 dB of high peak-signal-to-noise ratio (PSNR) and 1.6 to 2.0 of low M-singular value decomposition (M-SVD) values. In addition, the proposed approach outperforms conventional image watermarking approach in terms of normalized correlation (NC) values against several image processing attacks including noise addition, filtering, cropping, and JPEG compression.

• Journal of the Institute of Convergence Signal Processing
• /
• v.24 no.3
• /
• pp.160-165
• /
• 2023

In this study, we propose an SNS sentence writing assistance system that utilizes YOLO and GPT to assist users in writing texts with images, such as SNS. We utilize the YOLO model to extract objects from images inserted during writing, and also extract meta-information such as GPS information and creation time information, and use them as prompt values for GPT. To use the YOLO model, we trained it on form image data, and the mAP score of the model is about 0.25 on average. GPT was trained on 1,000 blog text data with the topic of 'restaurant reviews', and the model trained in this study was used to generate sentences with two types of keywords extracted from the images. A survey was conducted to evaluate the practicality of the generated sentences, and a closed-ended survey was conducted to clearly analyze the survey results. There were three evaluation items for the questionnaire by providing the inserted image and keyword sentences. The results showed that the keywords in the images generated meaningful sentences. Through this study, we found that the accuracy of image-based sentence generation depends on the relationship between image keywords and GPT learning contents.

• Journal of the Korean Society of Radiology
• /
• v.6 no.1
• /
• pp.5-10
• /
• 2012

There is a still unsettled issue about the comparison of long-term clinical effects between sirolimus-(SES) and paclitaxel-eluting stents (PES) for the patients with acute myocardial infarction (AMI). Therefore, we performed a retrospective analysis to evaluate the 4-year clinical outcome of SES as compared with PES after percutaneous coronary intervention (PCI) in patients with AMI. From January 2004 to August 2006, all consecutive patients with acute ST-segment elevation myocardial infarction (STEMI) underwent primary PCI and acute NSTEMI underwent PCI by implantation either SES or PES were enrolled. The occurrence of death, cardiac death, recurrent infarction, target vessel revascularization (TVR) and stent thrombosis were analyzed. The composite of major adverse cardiac events (MACE; death, recurrent infarction and TVR) were also analyzed. During the study period, total 668 AMI patients had visited. Of them, total 522 patients (299 with SES and 223 with PES) were enrolled. During 4-year clinical follow-up, there were similar occurrences of death ($18.3{\pm}3.0%$ vs. $14.6{\pm}2.2%$, p=0.26), cardiac death ($11.2{\pm}2.6%$ vs. $6.8{\pm}1.52%$, p=0.39), re-infarction ($6.4{\pm}1.8%$ vs. $3.3{\pm}1.1%$, p=0.31), and stent thrombosis ($5.4{\pm}1.7%$ vs. $3.2{\pm}1.1%$, p=0.53) between the two groups, consecutively. The occurrences of TVR ($10.0{\pm}3.0%$ vs. $4.0{\pm}1.2%$, p=0.008) and MACE ($29.4{\pm}3.5%$ vs. $19.4{\pm}2.5%$, p=0.003) were significantly higher in patients treated with PES than SES. In AMI patients treated with either SES or PES implantation, SES had a significantly lower risk of TVR and MACE during 4-year clinical follow-up. Rates of death, cardiac death or recurrent infarction, and stent thrombosis were similar.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.13 no.2
• /
• pp.233-242
• /
• 2009

This paper intend to show the affect of the intra-prediction on the typical digital watermarking method and the fact that the watermarking method has very low effectiveness when it is performed for the intra-frames of H.264. The target watermarking method was the one for imperceptibility and robustness and was assumed to be performed during the intra-compression process by the H.264 technique. Also this method was assumed to insert watermark data and to extract it for certification if needed. The problem is that the resulting data from the re-engineering of the watermark insertion process to extract the watermark data is different from the one before. We experimentally showed that it stems from the intra-prediction itself. That is, we showed that the resulting image data from only compression without watermarking changes if it is re-compressed by the same conditions as the first compression and it is because the intra-prediction modes as well as the coefficient values change. Also, we applied one blind and one semi-blind watermarking methods to show that the typical attacks after watermarking makes this problem much more serious and lowers the effectiveness of the watermarking method dramatically. Therefore we concluded by considering the experimental data that a typical watermarking method which has been researched so far cannot guarantee the effectiveness of intra-frame watermarking and it is highly required to developed a new kind of methodologies.

• The KIPS Transactions:PartD
• /
• v.10D no.3
• /
• pp.459-470
• /
• 2003

It is required to research on high-dimensional index structures for efficiently retrieving high-dimensional data because an attribute vector in data warehousing and a feature vector in multimedia database have a characteristic of high-dimensional data. For this, many high-dimensional index structures have been proposed, but they have so called ‘dimensional curse’ problem that retrieval performance is extremely decreased as the dimensionality is increased. To solve the problem, the cell-based filtering (CBF) scheme has been proposed. But the CBF scheme show a linear decreasing on performance as the dimensionality. To cope with the problem, it is necessary to make use of parallel processing techniques. In this paper, we propose a parallel CBF scheme which uses a horizontally-partitioned technique as declustering. In order to maximize the retrieval performance of the proposed parallel CBF scheme, we construct our parallel CBF scheme under a SN (Shared Nothing) cluster architecture. In addition, we present a data insertion algorithm, a rage query processing one, and a k-NN query processing one which are suitable for the SN cluster architecture. Finally, we show that our parallel CBF scheme achieves better retrieval performance in proportion to the number of servers in the SN cluster architecture, compared with the conventional CBF scheme.

• The Journal of Korea Institute of Information, Electronics, and Communication Technology
• /
• v.8 no.6
• /
• pp.460-469
• /
• 2015

This study's aim is for reducing big data transfer time of IoT healthcare devices by modulating digital bits into Manchester code including zero-voltage idle as information for secondary compressed binary cluster's compartment after two step compression of compressing binary data into primary and secondary binary compressed clusters for each binary clusters having compression benefit of 1 bit or 2 bits. Also this study proposed that as department information of compressed binary clusters, inserting idle signal into Manchester code will have benefit of reducing transfer time in case of compressing binary cluster into secondary compressed binary cluster by 2 bits, because in spite of cost of 1 clock idle, another 1 bit benefit can play a role of reducing 1 clock transfer time. Idle signal is also never consecutive because the signal is for compartment information between two adjacent secondary compressed binary cluster. Voltage transition on basic rule of Manchester code is remaining while inserting idle signal, so DC balance can be guaranteed. This study's simulation result said that even compressed binary data by another compression algorithms could be transferred faster by as much as about 12.6 percents if using this method.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.11 no.3
• /
• pp.880-890
• /
• 2010

The intrasvascular ultrasound (IVUS) imaging technique is used to diagnose cerebrovascular diseases such as stroke. Recently, elasticity imaging methods have been investigated to diagnose blood clots attached to blood vessel intima. However, the IVUS imaging technique is an invasive method that requires a transducer to be inserted into blood vessel. In this paper, strain images are obtained of blood clots attached to blood vessel intima with data acquired from outside the blood vessel using a linear array transducer. In order to measure the displacement of blood vessel accurately, experimental data are acquired by steering ultrasound beams so that they can intersect the blood vessel wall at right angles. The acquired rf data are demodulated to the baseband. The resulting complex baseband signals are then processed by an autocorrelation algorithm to compute the blood vessel movement and thereby produce strain image. This proposed method is verified by experiments on a plastic blood vessel mimicking phantom. The efficacy of the proposed method was verified using a home-made blood vessel mimicking phantom. The blood vessel mimicking phantom was constructed by making a 6 mm diameter hollow cylinder inside it to simulate a blood vessel and adhering 2 mm thick soft plaque to the inner wall of the hollow cylinder. The RF data were acquired using a clinical ultrasound scanner (Accuvix XQ, Medison, Seoul. Korea) with a 7.5 MHz linear array transducer by steering ultrasound beams in steps of $1^{\circ}$ from $-40^{\circ}$ to $40^{\circ}$ for a total of 81 angles. Experimental results show that the plaque region near the blood vessel wall is softer than background tissue. Although the imaging region is restricted due to the limited range of angles for which scan lines are perpendicular to the wall, the feasibility of strain imaging is demonstrated.

• Korean Journal of Remote Sensing
• /
• v.24 no.2
• /
• pp.99-106
• /
• 2008

A synthetic aperture radar (SAR) system is of great interest in many fields of civil and military applications because of all-weather and luminance free imaging capability. SAR image quality parameters such as spatial resolution, peak to sidelobe ratio (PSLR), and integrated sidelobe ratio (ISLR) can be normally estimated by modeling of impulse response function (IRF) which is obtained from various system design parameters such as altitude, operational frequency, PRF, etc. In modeling of IRF, however, background clutter environment surrounding the IRF is generally neglected. In this paper, analysis method for SAR mage quality is proposed in the real background clutter environment. First of all, SAR raw data of a point scatterer is generated based on various system parameters. Secondly, the generated raw data can be focused to ideal IRF by range Doppler algorithm (RDA). Finally, background clutter obtained from image of currently operating SAR system is applied to IRF. In addition, image quality is precisely analyzed by zooming and interpolation method for effective extraction of IRF, and then the effect of proposed methodology is presented with several simulation results under the assumption of estimation error of Doppler rate.