Originated from: an image edge detection based on data fusion and wavelet transform ... Journal of China University of Science and Technology 200 1 Wu,,.
Abstract: A method of image edge detection based on data fusion and wavelet transform is proposed. Firstly, the multi-spectral images in the same area are fused by wavelet, and then the edge discrimination threshold is dynamically adjusted by wavelet transform coefficients to detect the edge of the fused image. Experimental results show that this method can not only effectively suppress noise, but also has good adaptability to images with various edge features. This series of approximations has different resolutions, so it is called multiresolution analysis. On the basis of pyramid algorithm, continuous wavelet theory is extended to discrete field. From the concept of filter, wavelet transform is to filter the pleasure signal f(t) continuously with two sets of orthogonal Qualcomm and low-pass filters.
From: A Distortion Controllable Image Coding Method Radio Communication Technology 1997 Xu Peixia, Sun Gongxian.
Abstract: An arbitrary distortion image coding method based on wavelet transform and error feedback is proposed, which is suitable for remote database query and layered transmission of variable bit rate images. It decomposes the image into different resolutions through wavelet transform, and then gradually compensates by error feedback. Since coding errors of all previous resolutions can be compensated, undistorted images can be restored. It is also called multi-resolution analysis because it analyzes the location of signals on many different scales [2, 3]. Wavelet analysis covers a wide range, including: numerical analysis in the field of mathematics, constructing fast numerical methods, constructing curves and surfaces, solving differential equations, cybernetics and so on.
Source: Wavelet analysis of strain in the forestomach of ruminants, Journal of Xinjiang Agricultural University, 2003, Liu Haosen, Wei Junzhi.
Using DASP wavelet analysis module, the time-domain curves of contraction and relaxation strain of ruminants (sheep and cattle) in four physiological states (feeding, eating, ruminating and normal) were decomposed by wavelet, and the statistics of harmonic relaxation and contraction strain in different frequency bands were given. The results of wavelet analysis show that the main frequency and frequency spectrum of each measuring point in four physiological States are analyzed. This step-by-step analysis method is called multi-resolution analysis, which is an important direction of wavelet transform in practical engineering application. ξi is usually exponential, lognormal, normal and gamma distribution.
From: Application of Wavelet and Chaotic Learning Neural Network in Short-term Power Load Forecasting ... Computer Engineering and Application, 2003, Yang,,, Zheng Gang.
Abstract: A hybrid model of wavelet and neural network for short-term load forecasting of power system is proposed. Firstly, based on wavelet multi-resolution analysis method, the load sequence is decomposed into sequences with different frequency characteristics. Then, according to the characteristics of each component after decomposition, different neural network models are constructed to predict each component respectively. The neural network algorithm adopts chaotic learning algorithm. Compared with the traditional BP algorithm, this algorithm makes the system jump out of the shackles of local extremum and seek global optimization by using the swimming of chaotic orbit, which overcomes the essential problems of BP learning algorithm, accelerates the learning speed of the network and improves the learning accuracy. Finally, the prediction signal of each component is reconstructed to get the final prediction result. When constructing the network model, this paper considers the influence of climate factors as a set of input points of the network. Experimental results show that the load forecasting system based on this method has good accuracy and stability. After the iterative decomposition of LL(x, y), the multi-level decomposition of two-dimensional image f(x, y), or multi-resolution analysis, is obtained. The result of wavelet transform is multiple high-frequency band data and one low-frequency band data of the original signal in a series of octave-divided frequency bands.
Source: research on image compression algorithm based on statistical characteristics of wavelet transform Wu Baoming, Peng, 2002 Journal of Biomedical Engineering.
Abstract: Statistical distribution of image energy is an important basis of image compression processing. On the basis of studying the statistical characteristics of wavelet subband images, a new image quantization coding algorithm based on the statistical characteristics of wavelet subband images and human visual characteristics is proposed. Experiments show that the algorithm has the characteristics of simple calculation and high compression efficiency. In this sense, wavelet analysis can also be called multi-resolution analysis, which is a milestone in the history of Fourier analysis. It has been widely used in signal processing, seismic exploration, celestial body identification, mechanical fault diagnosis and monitoring and other scientific and technological fields.
From: ECG signal processing technology and wavelet transform method Journal of Dalian Institute of Light Industry 200 1 Zhang Shuqing, Li Changwu, Wang Li.
Abstract: Two methods of ECG signal processing are given. The first method is to use synthesis technology, which can ensure the integrity of the waveform and is easy to implement. The second method is to use wavelet analysis. Wavelet transform is suitable for nonstationary signal analysis, ECG data preprocessing and feature extraction. In this paper, Mallat algorithm is used to decompose ECG signals at multiple scales, which is called multi-resolution analysis. The mode of fingerprint image is quasi-periodic mode, and the ridge direction and spatial frequency of different regions represent the essential attributes of different fingerprint images.
From: Fast Fingerprint Identification Based on Algebraic and Geometric Features Journal of Zhejiang Sci-Tech University Michelle Li and Hu Zhihui, 2005.
Abstract: In order to learn from each other in the matching method based on geometric features of fingerprints, a new phased matching method based on algebraic features and geometric features of fingerprints is proposed. Experiments show that the matching time is reduced by 47.5% while ensuring a high recognition rate. The algorithm is expected to develop into a practical and effective fingerprint identification technology. This kind of analysis of things from coarse to fine is called multi-resolution analysis. In time domain, the scale changes from large to small, and the corresponding frequency domain scale changes from small to large. Low-pass filter can get large-scale information, that is, low-frequency information-signal contour information, while high-pass filter can get small-scale information, that is, high-frequency information-noise and mutation information.
Source: ECG signal noise processing based on wavelet transform Journal of Northwestern Polytechnical University 2005 Zhang Jingzhou, Shou Guofa, Dai Guanzhong.
Based on the multi-resolution analysis of wavelet transform, through the analysis of ECG signal and its noise, different wavelet denoising algorithms are designed for baseline drift, power frequency interference and EMG interference in ECG signal. The ECG signal in MIT /BIH international standard database and the ECG signal generated by program simulation are used to simulate and verify the algorithm respectively. The results show that the algorithm can effectively filter out several main noises in ECG signal detection with little distortion, and can meet the clinical requirements for ECG signal waveform analysis and diagnosis. The basic idea is to regard the function f in L ~ 2 (r) space as the limit of gradual approximation. Each approximation is a smooth version of f, and the resolution of successive approximation is different, so it is called multi-resolution analysis. A progressively approaching frame needs some translation invariance. More precisely, multiresolution analysis is nested.
Derived from: wavelet basis of standard orthogonal compact support set and decomposition of seismic data ...
Abstract: This paper discusses the orthogonal compactly supported wavelet bases of Daubechies standard. With the help of multi-resolution analysis method, the decomposition and reconstruction algorithm of seismic data is established to compress and reconstruct the measured seismic data. This nested structure is usually called multi-resolution analysis. It is worth noting that subspace,. Z can't be obtained by integer translation of a single function, which is one of the important characteristics of multiwavelets different from traditional wavelets.
From: Research progress of multiwavelet and its application in power system ... Power system automation, 2004.
Abstract: Multiwavelets can have the properties of symmetry, orthogonality, short support and high-order vanishing moment at the same time, which is incomparable to traditional wavelets. By introducing the earliest multiwavelets, the basic properties of multiwavelets are introduced. The research status of multiwavelet theory is discussed in detail, and several commonly used multiwavelets are compared. The pretreatment of multiwavelet is discussed and analyzed in depth, and classified. Combined with the field of power system, the problems existing in the practical application of multiwavelet theory are put forward and discussed. Finally, the future research problems of multiwavelet and its application in power system are prospected.