Earticle

본 논문에서는 가중 퍼지소속함수 기반 신경망(neural network with weighted fuzzy membership functions; NEWFM) 기반의 웨이블릿 변환(wavelet transform)과 힐버트 변환(Hilbert transform)에 의해 추출한 첨점(peak)을 사용하여 뇌파(EEG)로부터 정상 파형과 간질 파형을 분류하는 새로운 방안을 제안하였다. NEWFM의 입력을 추출하는데 다음과 같은 3개의 단계가 수행되었다. 첫 번째 단계에서는 뇌파로부터 잡음을 제거하기 위해서 웨이블릿 변환을 사용하였다. 두 번째 단계에서는 웨이블릿 계수로부터 첨점(peak)을 추출하기 위해서 힐버트 변환을 사용하였다. 또한 크기가 큰 첨점을 추출하기 위해서 첨점의 평균값보다 큰 첨점만을 선택하였다. 세 번째 단계에서는 통계적 방법을 이용하여 첨점으로부터 NEWFM의 입력으로 사용할 16개의 특징을 추출하였다. NEWFM은 이들 16개의 특징을 입력으로 사용하여 99.25%, 99.4%, 99%의 정확도, 특이도, 민감도를 각각 구하였다. 향후 연구에서는 특징선택을 이용하여 16개의 특징으로부터 좋은 특징을 선택하여 정확도를 향상시킬 계획이다.

This study proposed new methods to classify normal and epileptic seizure signals from EEG signals using peaks extracted by wavelet transform(WT) and Hilbert transform(HT) based on a neural network with weighted fuzzy membership functions(NEWFM). This study has the following three steps for extracting inputs for NEWFM. In the first step, the WT was used to remove noise from EEG signals. In the second step, the HT was used to extract peaks from the wavelet coefficients. We also selected the peaks bigger than the average of peaks to extract big peaks. In the third step, statistical methods were used to extract 16 features used as inputs for NEWFM from peaks. The proposed methodology shows that accuracy, specificity, and sensitivity are 99.25%, 99.4%, 99% with 16 features, respectively. Improvement in feature selection method in view to enhancing the accuracy is planned as the future work for selecting good features from 16 features.

본 논문에서는 신호 처리 기술과 가중 퍼지소속함수 기반 신경망 (Neural Network with Weighted Fuzzy Membership Functions; NEWFM)을 이용하여 간질을 검출하는 방안을 제안하였다. 신호 처리 기술로는 웨이블릿 변환(Wavelet Transform), 점증적 증가 방법, 위상공간 재구성(Phase Space Reconstruction)을 이용하였다. 신호 처리 기술의 첫 번째 단계에서는 웨이블릿 변환을 이용하여 뇌파로부터 웨이블릿 계수를 추출하였다. 두 번째 단계에서는 점증적 증가 방법을 이용하여 웨이블릿 계수로부터 첨점(Peak)을 추출하였다. 세 번째 단계에서는 위상공간 재구성을 이용하여 추출된 첨점으로부터 3차원 다이어그램을 생성하였다. NEWFM의 입력으로 사용할 16개의 특징을 추출하기 위하여 유클리드 거리와 통계적 방법을 이용하였다. 이들 16개의 특징을 NEWFM의 입력으로 사용하여 97.5%, 100%, 95%의 정확도, 특이도, 민감도를 각각 구하였다.

This study proposed signal processing techniques and neural network with weighted fuzzy membership functions(NEWFM) to detect epileptic seizure from EEG signals. This study used wavelet transform(WT), sequential increment method, and phase space reconstruction(PSR) as signal processing techniques. In the first step of signal processing techniques, wavelet coefficients were extracted from EEG signals using the WT. In the second step, sequential increment method was used to extract peaks from the wavelet coefficients. In the third step, 3D diagram was produced from the extracted peaks using the PSR. The Euclidean distances and statistical methods were used to extract 16 features used as inputs for NEWFM. The proposed methodology shows that accuracy, specificity, and sensitivity are 97.5%, 100%, 95% with 16 features, respectively.

특징 선택은 중복 또는 서로간의 관련이 없는 특징을 제거하여 분류 성능을 향상시키는 기술이다. 본 논문 에서는 가중 퍼지소속함수 기반 신경망 (Neural Network with Weighted Fuzzy Membership Functions; NEWFM)에서 제공하는 가중 퍼지소속함수의 경계합 (Bounded Sum of Weighted Fuzzy Membership functions, BSWFM)의 무게중 심간의 거리를 이용한 새로운 특징 선택을 제안하여 분류 성능을 향상시켰다. 이러한 거리 기반의 특징 선택을 이용 하여 초기 24개의 특징으로부터 무게중심간의 거리가 짧은 특징을 하나씩 제거되면서 분류 성능이 가능 높은 22개 의 최소 특징을 선택하였다. 이들 22개의 최소 특징을 NEWFM의 입력으로 사용하여 97.7%, 99.7%, 98.7%의 민감 도, 특이도, 정확도를 각각 구하였다.

Feature selection is the technique to improve the classification performance by using a minimal set by removing features that are not related with each other and characterized by redundancy. This study proposed new feature selection using the distance between the center of gravity of the bounded sum of weighted fuzzy membership functions (BSWFMs) provided by the neural network with weighted fuzzy membership functions (NEWFM) in order to improve the classification performance. The distance-based feature selection selects the minimum features by removing the worst features with the shortest distance between the center of gravity of BSWFMs from the 24 initial features one by one, and then 22 minimum features are selected with the highest performance result. The proposed methodology shows that sensitivity, specificity, and accuracy are 97.7%, 99.7%, and 98.7% with 22 minimum features, respectively.

A hybrid microcalcification detection method based on Fisher discriminant and SVM was presented because signal-to-noise ratio of mammogram image was very low, and microcalcifications were very small and their shape was irregular. Firstly, low frequency information of tissue was removed by wavelet transform in order to reduce the tissue effect to microcalcification segment. Secondly, Fisher discriminant was adopted to find optimum threshold, meanwhile microcalcification was segmented. Lastly, SVM classifier was adopted to recognize true microcalcifications. Experiment results showed that Fisher discriminant could validly segment microcalcifications and the number of false positive targets was less than OSTU’s. Detection ratio of our algorithm was about 97%.

Multispectral approach to brain MRI analysis has shown great advance recently in pathology and tissue analysis. However, poor performance of the feature extraction and classification techniques involved in it discourages radiologists to use it in clinical applications. Transform based feature extraction methods like Independent Component Analysis (ICA) and its variants have contributed a lot in this research field. But these global transforms often fails in extraction of local features like small lesions from clinical cases and noisy data. Feature extraction part of the recently introduced Multiresolution Independent Component Analysis (MICA) algorithm in microarray classification is proposed in this work to resolve this issue. Effectiveness of the algorithm in MRI analysis is demonstrated by training and classification with Support Vector Machines (SVM). Both synthetic and real abnormal data from T1-weighted, T2-weighted, proton density, fluid-attenuated inversion recovery and diffusion weighted MRI sequences are considered for detailed evaluation of the method. Tanimoto index, sensitivity, specificity and accuracy of the classified results are measured and analyzed for brain abnormalities, affected white matter and gray matter tissues in all cases including noisy environment. A detailed comparative study of classification using MICA and ICA is also carried out to confirm the positive effect of the proposed method. MICA based SVM is found to yield very good results in anomaly detection, around 2.5 times improvement in classification accuracy is observed for abnormal data analysis.

In this paper a watermarking technique using hybrid wavelet transforms obtained from sinusoidal and non-sinusoidal component orthogonal transforms is proposed. Sinusoidal transform DCT and non-sinusoidal transforms Walsh, Haar and Discrete Kekre Transform are used to generate hybrid wavelet transforms namely DCT-Walsh, Walsh-DCT, DCT-Haar, Haar-DCT, DCT-DKT and DKT-DCT. Size of each component transform matrix is varied suitably from 4, 8, 16, 32, and 64 to generate hybrid wavelet transform matrix for host and watermark. The best size combination is further applied column wise and row wise to host and watermark and to embed the watermark middle frequency regions of host is selected. Embedding is first done without sorting the hybrid wavelet transform coefficients of host and watermark and then sorting is applied to observe the difference in the achieved robustness. Performance of proposed technique is evaluated against various attacks to decide whether sinusoidal transform when used as base transform matrix or local transform matrix is more robust.

The image fusion is the process of combining relevant information from two or more images into a single image. The resulting image will be more informative than any of the input images. A new approach for object extraction from high-resolution images is presented in this report. In this paper, we have presented image fusion based on wavelet transform.

EEG (Electroencephalograph) is a technique for identifying neurological disorders. There are arious neurological disorders like Epilepsy, brain cancer, etc. Epilepsy is one of the common eurological disorders. In this paper we propose a technique of detecting epilepsy disorder using iscrete wavelet transform using MATLAB. The back propagation algorithm is also used in the lassification network. This paper also provides a technique of detecting epilepsy disorder with reat accuracy.

과표본화 이산 웨이브렛 변환의 가장 대표적으로 응용되는 분야는 디지털 영상에 존재하는 잡음을 제거하는 기술이 다. 이중 밀도 이산 웨이브렛 변환을 이중 트리 이산 웨이브렛 변환과 비교하면, 거의 유사한 특징을 가진다. 본 논문에서는 잡음이 포함된 디지털 영상에 여러 이산 웨이브렛 변환들을 수행하고 생성된 부대역에 임계값 처리 기법을 적용하여 잡음을 제거한 다음 복원한 영상의 성능을 평가하는 실험을 수행하였다. 적당한 임계값을 설정하여 효과적인 잡음제거가 가능하다. 본 논문에서는 여러 방법의 실험 결과에서 제안하는 3방향 분리처리 2차원 이중 밀도 이산 웨이브렛 변환 방법이 우수하다는 것을 확인할 수 있었다.

The standard application area of over-sampled discrete wavelet transform is noise removal technology for digital images. Comparing dual density discrete wavelet transform with dual tree discrete wavelet transform, we have almost similar characteristics. In this paper, several discrete wavelet transforms are accomplished on digital image existing with noise, noises are removed with threshold processing algorithm on subband, performance evaluation experiments of the reconstructed images are accomplished. If we decide appropriate threshold value, the effect noise removal is possible. In this paper, we can certified that the suggested algorithm of 3-direction separable processing with 2 dimension dual density discrete wavelet transform is superior to several experiment results

An application of wavelet analysis to power system transient generated signals is presented in this paper. With the time-frequency localisation characteristics embedded in wavelets, the time and frequency information of a waveform can be presented as a visualised scheme. This feature is very important for non-stationary signals analysis such as the ones generated from power system disturbances. Unlike the Fourier transform, the wavelet transform approach is more efficient in monitoring fault signals as time varies. For time intervals where the function changes rapidly, this method can zoom in on the area of interest for better visualisation of signal characteristics.

Existence of spectral density of wavelet transform in the case of fractional Brownian motion is proved by Kato and Masry (1999). Given their results, we provide supplementary results on spectral behavior at thezero frequency. It is found that the spectral density at the zero frequency, determined by the memory parameter and the number of vanishing moments of the wavelets, generates possible degeneracy. Our results can be understood as spectral version of decorrelation properties of wavelet transforms.

A novel method for evaluating the anisotropy of the deterministic features in a stochastic 2D data is introduced. The ability of the wavelet transform for the identification of the abrupt discontinuities could be used to characterize the boundary of the deterministic area in a 2D stochastic data, such as flocs in paper structure. The one-dimensional wavelet transform with a small-scale range in MD and CD could quantify the amount of the edge in both directions, depending on the intensity of each floc. The flocs that are aligned in the MD direction result in a higher value of local wavelet energy in the CD direction. Therefore, the ratio of the total wavelet energy in CD and MD directions can be used as a new anisotropy index. This index is a measure of the floc-orientation and can provide an excellent tool to obtain the orientation distribution and the major oriented angle of flocs. Various simulated images and real stochastic data such as local gloss variation of printed image and formation image, have been tested and the results show this analysis method is very reliable to measure the anisotropy of the deterministic features.

In this study the complex-valued continuous wavelet transform (CWT) has been applied in detection of Electrocardiograms (ECG) as response to various signal classification methods such as Fourier transforms and other tools of time frequency analysis. Experiments have shown that CWT may serve as a detector of non-stationary signal changes as ECG. The tested signal is corrupted by short time events. We applied CWT to detect short-time event and the result image representation of the signal has showed us that one can easily find the discontinuity at the time scale representation. Analysis of ECG signal using complex-valued continuous wavelet transform is the first step to detect possible changes and alternans. In the second step, modulus and phase must be thoroughly examined. Thus, short time events in the ECG signal, and other important characteristic points such as frequency overlapping, wave onsets/offsets extrema and discontinuities even inflection points are found to be detectable. We have proved that the complex-valued CWT can be used as a powerful detector in ECG signal analysis.

This paper presents an algorithm based on a combination of Discrete Wavelet Transforms and neural networks for detection and classification of internal faults in a two-winding three-phase transformer. Fault conditions of the transformer are simulated using ATP/EMTP in order to obtain current signals. The training process for the neural network and fault diagnosis decision are implemented using toolboxes on MATLAB/Simulink. Various cases and fault types based on Thailand electricity transmission and distribution systems are studied to verify the validity of the algorithm. It is found that the proposed method gives a satisfactory accuracy, and will be particularly useful in a development of a modern differential relay for a transformer protection scheme.

One of the main techniques or diagnosing heart disease is by examining the electrocardiogram(ECG). Many studies on detecting the QRS complex, P, and T waves have been performed because meaningful information is contained in these parameters. However, the earlier detecting techniques can not effectively extract those parameters from the ECG that is severely contaminated by noise source such 60Hz powerline interference, motion artifact and baseline drift. in this paper, we performed the extracting parameters from and recovering the ECG signal using wavelet transform that has recently been applying to various fields.

볼륨 데이터는 시뮬레이션 통해 생성되거나 고성능 측정 장비를 이용해 측정된 값으로 구성되는 고차원 데이터의 한 형태로서 다양한 자연과학과 공학분야에서 폭넓게 활용되고 있다. 최근에는 각 분야에서 생성되는 계산 데이터의 용량이 점점 더 증가하고 있기 때문에 이런 대용량의 볼륨 데이터를 효과적으로 처리하기 위한 기법들에 관한 연구가 수행되고 있으며, 특히 대용량 볼륨 데이터 압축 기법에 대한 필요성이 증가하고 있다. 본 논문에서는 Daubechies 웨이블릿 변환과 zerobit 인코딩 스킴을 응용한 새로운 볼륨 데이터 압축 기법을 제안한다. 이 방법은 기존의 압축 방법에 비해 복원 데이터의 손실이 낮기 때문에 정밀한 영상을 요구하는 대용량 데이터 압축에 유용하게 사용될 수 있다.

Parameter estimation for the probability model of wavelet coefficients is essential to the wavelet-domain interpolation. However, phase uncertainty, one well-known drawback of the orthogonal wavelet transforms, make it difficult to estimate parameters. In this paper, we exploit a phase shifting matrix in order to improve the accuracy of estimation. Nonlinear modeling to capture the interscale characteristics is also described. The experimental results show that the proposed method outperforms the previous wavelet-domain interpolation method as well as the conventional bicubic method.

목적: 다른 생체신호와 마찬가지로 망막전도(electroretinogram, ERG) 신호도 측정시 잡음이 발생한다. 이 잡음을 효과적으로 제거하여 망막관련 진단의 정확도를 높이고자 하였다. 방법: ERG 신호에 60 Hz 잡음과 백색잡음을 발생시켜 샘플링 신호를 만들었다. 웨이브렛 변환과 대역통과 필터를 이용하여 잡음를 제거하였다. 푸리에 변환 스펙트럼을 이용하여 제거된 주파수를 비교하였다. 신호대잡음비(signal to noise ratio, SNR)를 이용하여 제거된 잡음을 수치적으로 비교하였다. 결과: 푸리에 변환 스펙트럼을 비교한 결과 웨이브렛 변환에서는 60 Hz 잡음은 완전히 제거 되었으며 백색잡음도 많이 제거되었다. 대역통과필터에서는 60 Hz와 백색잡음 남아 있었다. 신호대잡음비를 비교한 결과에서는 웨이브렛 변환은 22.8638, 대역통과 필터는 4.0961로 나타났다. 결론: 웨이브렛 변환을 이용하여 잡음 제거시 신호의 왜곡을 적게 발생시켜 제거할 수 있었다. 망막전도 신호를 이용한 망막 진단에 정확도를 높일 수 있을 것으로 기대된다.

Purpose: Electroretinogram(ERG) signal noise as well as conducting other bio-signal measurement were generated. It was intened to enhance the accuracy of retinal-related diagnosis with removing signal noise. Methods: Sampling signal was made with generating 60 Hz noise and white noise. The noise were removed using wavelet transforms and bandpass filter. De-noising frequency was compared with Fourier transform spectrum. Removed noises were compared numerically using SNR(signal to noise ratio). Results: The result compared Fourier transform spectrum was showed that 60 Hz noise removed completely and most of white noise was removed by wavelet transforms. 60 Hz and the white noise remained using bandpass filters. The result compared SNR showed that wavelet transforms was 22.8638 and bandpass filter was 4.0961. Conclusions: Wavelet transform showed less signal distortion in removing noise. ERG signal is expected to improve the accuracy of retinal-related diagnosis.

본 논문에서 사용한 웨이블릿 변환의 기저 함수는 일반적인 웨이블릿 변환과 다른 리프팅 스킴을 사용하여 만들어 졌다. 리프팅 스킴은 푸리에 변환을 사용하여 기저 함수를 생성하지 않는 새로운 쌍직교 웨이블릿 기저 함수를 생성하는 방법이다 본 본문은 리프팅 스킴을 이용한 새로운 영상 압축 및 에지 검출 방법을 제안하고 있다. 그리고 이 방법은 부분 복원과 공간 복원을 할 수 있어 데이터 가시화를 향상시킬 수 있다. 다양한 해상도에서의 근사 영상은 원래 영상으로부터 적은 정보만으로 다양한 크기의 특징을 뽑아낼 수 있고, 적은 양의 스케일링 계수를 사용하여 생성된 근사 영상은 빠르게 원래 영상의 대략적인 개요만이 필요할 때 유용하게 사용된다. 본 논문에서 제안한 영상 압축 및 에지 검출 기법은 멀티미디어 데이터베이스에서 데이터 관리와 데이터 가시화를 향상시킬 수 있는 좋은 기틀을 마련해 준다.

The basis function of wavelet transform used in this paper is constructed by using lifting scheme, which is different from general wavelet transform. Lifting scheme is a new biorthogonal wavelet con-structing method, that does not use Fourier transform for constructing its basis function. In this paper, an image compression and reconstruction method using the lifting scheme was proposed. And this method improves data visualization by supporting a partial reconstruction and a local reconstruction. Approx- imations at various resolutions allow extracting various sizes of feature from an image or signal with a small amount of original information. An approximation with small size of scaling coefficients gives a brief outline of features at fast. Image compression and edge detection techniques provide good frame- works for data management and visualization in multimedia database.

Recently, the wavelet transform has been a new and powerful tool for signal processing. It is more suitable specially for the feature extraction and detection of non-stationary signals than traditional methods such as, the Fourier Transform(FT), the Fast Fourier Transform(FFT) and the Least Square Method etc. because of the characteristic of the multi-scale analysis and time-frequency domain localization. The wavelet transform has been developed for the analysis of PD pulse signal to raise in the progress of insulation degradation. In this paper, the wavelet transform was applied to one foundational method for feature extraction. For the obtain experimental data, a computer-aided partial discharge measurement system with a single acoustic sensor was used. If we are applying to the neural network method the accumulated data through the extracted feature, it is expected that we can detect the PD pulse signal in the insulation materials on the on-line.