• The Transactions of the Korea Information Processing Society
• /
• v.2 no.6
• /
• pp.903-917
• /
• 1995

In most database systems images have been indexed indirectly using related texts such as captions, annotations and image attributes. But there has been an increasing requirement for the image database system supporting the storage and retrieval of images directly by content using the information contained in the images. There has been a few indexing methods based on contents. Among them, Pertains proposed an image indexing method considering spatial relationships and properties of objects forming the images. This is the expansion of the other studies based on '2-D string. But this method needs too much storage space and lacks flexibility. In this paper, we propose a more flexible index structure based on kd-tree using paging techniques. We show an example of extracting keys using normalization from the from the raw image. Simulation results show that our method improves in flexibility and needs much less storage space.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.15 no.1
• /
• pp.133-138
• /
• 2020

This paper introduces a method for improving accuracy using the first convolution layer, which is not used in most modified CNN(: Convolution Neural Networks). In CNN, such as GoogLeNet and DenseNet, the first convolution layer uses only the traditional methods(3×3 convolutional computation, batch normalization, and activation functions), replacing this with RGB-csb. In addition to the results of preceding studies that can improve accuracy by applying RGB values to feature maps, the accuracy is compared with existing CNN using a limited number of images. The method proposed in this paper shows that the smaller the number of images, the greater the learning accuracy deviation, the more unstable, but the higher the accuracy on average compared to the existing CNN. As the number of images increases, the difference in accuracy between the existing CNN and the proposed method decreases, and the proposed method does not seem to have a significant effect.

• Journal of the Institute of Electronics Engineers of Korea SP
• /
• v.41 no.4
• /
• pp.45-50
• /
• 2004

In this paper, we propose a new PCA-based LDA Mixture Algorithm(PLMA) for real-time face recognition system. This system greatly consists of the two parts: 1) face extraction part; 2) face recognition part. In the face extraction part we applied subtraction image, color filtering, eyes and mouth region detection, and normalization method, and in the face recognition part we used the method mixing PCA and LDA in extracted face candidate region images. The existing recognition system using only PCA showed low recognition rates, and it is hard in the recognition system using only LDA to apply LDA to the input images as it is when the number of image pixels ire small as compared with the training set. To overcome these shortcomings, we reduced dimension as we apply PCA to the normalized images, and apply LDA to the compressed images, therefore it is possible for us to do real-time recognition, and we are also capable of improving recognition rates. We have experimented using self-organized DAUface database to evaluate the performance of the proposed system. The experimental results show that the proposed method outperform PCA, LDA and ICA method within the framework of recognition accuracy.

• Journal of the Institute of Electronics Engineers of Korea SC
• /
• v.48 no.2
• /
• pp.80-85
• /
• 2011

Alzheimer's disease and Parkinson's disease are the most common neurodegenerative disorders. In this paper, we proposed a feature extraction method for classification of AD and PD based on micro-Raman spectra from platelet. The first step of the preprocessing is a simple smoothing followed by background elimination to the original spectra to make it easy to measure the intensity of the peaks. The last step of the preprocessing was peak alignment with the reference peak. After the inspection of the preprocessed spectra, we found that proportion of two peak intensity at 743 and $757cm^{-1}$ and peak intensity at 1248 and $1448cm^{-1}$ are the most discriminative features. Then we apply mapstd method for normalization. The method returned data with means to 0 and deviation to 1. With these three features, the classification result involving 263 spectra showed about 95.8% true classification in case of MAP(maximum a posteriori probability).

• The Journal of Korean Institute of Communications and Information Sciences
• /
• v.37 no.7A
• /
• pp.569-580
• /
• 2012

In this paper, an enhanced initial ranging algorithm for large-cell coverage fixed wireless communication system is proposed. In typical wireless communication system such as WiBro, because a round-trip delay between a transmitter and a receiver is within one OFDM (Orthogonal Frequency Division Multiplexing) symbol duration, a frequency-domain differential correlation method is generally used. However, the conventional method cannot be applied due to an increase of a maximum time delay in large-cell system. In case of an accumulative differential method, estimation errors can occur because of frequent sign transitions. In this paper, therefore, we propose an algorithm which can estimate a total timing offset in a ranging channel structure for 15 km cell. The proposed method can improve performance by sign comparison based sign error correction rule between the estimated values and using a weighting scheme based on channel correlation, the number of accumulations, and the noise reduction effect in normalization process. Also, it can estimate the integer timing offset of symbol duration by comparing peak-powers after compensating for the fractional timing offset of symbol duration.

• The Journal of the Acoustical Society of Korea
• /
• v.35 no.6
• /
• pp.501-509
• /
• 2016

We propose a method to improve the robustness of speaker verification on short test utterances. The accuracy of the state-of-the-art i-vector/probabilistic linear discriminant analysis systems can be degraded when testing utterance durations are short. The proposed method compensates for utterance variations of short test feature vectors using deep neural networks. We design three different types of DNN (Deep Neural Network) structures which are trained with different target output vectors. Each DNN is trained to minimize the discrepancy between the feed-forwarded output of a given short utterance feature and its original long utterance feature. We use short 2-10 s condition of the NIST (National Institute of Standards Technology, U.S.) 2008 SRE (Speaker Recognition Evaluation) corpus to evaluate the method. The experimental results show that the proposed method reduces the minimum detection cost relative to the baseline system.

• The Journal of the Korea institute of electronic communication sciences
• /
• v.10 no.12
• /
• pp.1329-1336
• /
• 2015

Due to the advances in hardware technologies for low-power CMOS cameras, there have been various researches on wireless video sensor network(WVSN) applications including agricultural monitoring and environmental tracking. In such a system, its core technologies include video compression and wireless transmission. Since data of video sensors are bigger than those of other sensors, it is particularly necessary to estimate precisely the traffic after video encoding. In this paper, we present an estimation method for the encoded video traffic in WVSN networks. To estimate traffic characteristics accurately, the proposed method first measures complexities of frames and then applies them to the bit rate estimation adaptively. It is shown by experimental results that the proposed method improves the estimation of bit rate characteristics by more than 12% as compared to the existing method.

• Journal of the Korea Academia-Industrial cooperation Society
• /
• v.12 no.9
• /
• pp.4119-4124
• /
• 2011

In a MPC coding using excitation source of voiced and unvoiced, it would be a distortion of speech waveform in case of increasing or decreasing of speech signal amplitude in a frame. This is caused by normalization of synthesis speech signal in the process of restoration the multi-pulses of representation section. To solve this problem, this paper present a method of amplitude compensation(AC-MPC) in a multi-pulses each pitch interval in order to reduce distortion of speech waveform. I was confirmed that the method can be synthesized close to the original speech waveform. And I evaluate the MPC and AC-MPC using amplitude compensation method. As a result, SNRseg of AC-MPC was improved 0.7dB for female voice and 0.7dB for male voice respectively. Compared to the MPC, SNRseg of AC-MPC has been improved that I was able to control the distortion of the speech waveform finally. And so, I expect to be able to this method for cellular phone and smart phone using excitation source of low bit rate.

• Journal of the Korea Institute of Information and Communication Engineering
• /
• v.12 no.2
• /
• pp.315-320
• /
• 2008

Nowadays, a lot of related data obtained from these research could be given a new present meaning to accomplish the original purpose of the whole research as a human genome project. In such a thread, construction of gene expression analysis system and a basis rank analysis system is being watched newly. Recently, being identified fact that particular sub-class of tumor be related with particular chromosome, microarray started to be used in diagnosis field by doing cancer classification and predication based on gene expression information. In this thesis, we used cDNA microarrays of 3840 genes obtained from neuronal differentiation experiment of cortical stem cells on white mouse with cancer, created system that can extract informative gene list through normalization separately and proposed combination method for selecting more significant genes. And possibility of proposed system and method is verified through experiment. That result is that PC-ED combination represent 98.74% accurate and 0.04% MSE, which show that it improve classification performance than case to experiment after generating gene list using single similarity scale.

• KSII Transactions on Internet and Information Systems (TIIS)
• /
• v.14 no.9
• /
• pp.3712-3729
• /
• 2020

Smoke detection is helpful for early fire detection. With its large coverage area and low cost, vision-based smoke detection technology is the main research direction of outdoor smoke detection. We propose a two-stage smoke detection method combined with block Deep Normalization and Convolutional Neural Network (DNCNN) and visual change image. In the first stage, each suspected smoke region is detected from each frame of the images by using block DNCNN. According to the physical characteristics of smoke diffusion, a concept of visual change image is put forward in this paper, which is constructed by the video motion change state of the suspected smoke regions, and can describe the physical diffusion characteristics of smoke in the time and space domains. In the second stage, the Support Vector Machine (SVM) classifier is used to classify the Histogram of Oriented Gradients (HOG) features of visual change images of the suspected smoke regions, in this way to reduce the false alarm caused by the smoke-like objects such as cloud and fog. Simulation experiments are carried out on two public datasets of smoke. Results show that the accuracy and recall rate of smoke detection are high, and the false alarm rate is much lower than that of other comparison methods.