32 results about "Multiresolution analysis" patented technology

A system and method for identifying critical features in an ordered scale space within a multi-dimensional feature space is described. Features are extracted from a plurality of data collections. Each data collection is characterized by a collection of features semantically-related by a grammar. Each feature is normalized and frequencies of occurrence and co-occurrences for the feature for each of the data collections is determined. The occurrence frequencies and the co-occurrence frequencies for each of the features are mapped into a set of patterns of occurrence frequencies and a set of patterns of co-occurrence frequencies. The pattern for each data collection is selected and distance (similarity) measures between each occurrence frequency in the selected pattern is calculated. The occurrence frequencies are projected onto a one-dimensional document signal in order of relative decreasing similarity using the similarity measures. Wavelet and scaling coefficients are derived from the one-dimensional document signal using multiresolution analysis.

A new multiresolution analysis (wavelet) assisted reinforcement learning (RL) based control strategy that can effectively deal with both multiscale disturbances in processes and the lack of process models. The application of wavelet aided RL based controller represents a paradigm shift in the control of large scale stochastic dynamic systems of which the control problem is a subset. The control strategy is termed a WRL-RbR controller. The WRL-RbR controller is tested on a multiple-input-multiple-output (MIMO) Chemical Mechanical Planarization (CMP) process of wafer fabrication for which process model is available. Results show that the RL controller outperforms EWMA based controllers for low autocorrelation. The new controller also performs quite well for strongly autocorrelated processes for which the EWMA controllers are known to fail. Convergence analysis of the new breed of WRL-RbR controller is presented. Further enhancement of the controller to deal with model free processes and for inputs coming from spatially distributed environments are also addressed.

A new multiresolution analysis (wavelet) assisted reinforcement learning (RL) based control strategy that can effectively deal with both multiscale disturbances in processes and the lack of process models. The application of wavelet aided RL based controller represents a paradigm shift in the control of large scale stochastic dynamic systems of which the control problem is a subset. The control strategy is termed a WRL-RbR controller. The WRL-RbR controller is tested on a multiple-input-multiple-output (MIMO) Chemical Mechanical Planarization (CMP) process of wafer fabrication for which process model is available. Results show that the RL controller outperforms EWMA based controllers for low autocorrelation. The new controller also performs quite well for strongly autocorrelated processes for which the EWMA controllers are known to fail. Convergence analysis of the new breed of WRL-RbR controller is presented. Further enhancement of the controller to deal with model free processes and for inputs coming from spatially distributed environments are also addressed.

Function spaces defined by scaling functions are used to generate bandlimited noise octaves and other attribute data sets. Scaling functions are basis functions that admit multiresolution analysis and include piecewise constant scaling functions, piecewise polynomial scaling functions, bandlimited scaling functions, Daubeschies scaling functions, as well as other multiresolution analysis scaling basis functions known to those of skill in the art. Scaling basis functions can be locally supported or have infinite support. The properties of the scaling basis functions used to construct bandlimited noise octaves may ensure that any bandlimited noise octave at resolution level N is orthogonal to bandlimited noise octaves and their associated scaling basis functions at all resolution levels less than N. Bandlimited noise octaves can be scaled to any resolution level and guaranteed to have no effect on images at any lower resolution level.

The invention discloses a microseismic signal noise reduction filtering method based on a VMD and a wavelet packet and belongs to the signal processing technology field. In the invention, a mode of combining the VMD and the wavelet packet is adopted; adaptivity of a VMD decomposition method and characteristics of a high mathematical theory basis, high frequency noise suppression and the like in the algorithm are used; and the wavelet packet has the characteristics of carrying out multilevel division on a frequency band, further decomposing a high frequency portion which is not subdivided in multiresolution analysis, and adaptively selecting a frequency band according to a characteristic of an analyzed signal so that a time frequency resolution is increased. In the invention, based on a condition of maintaining randomness, non-stationary performance and a burst transient characteristic of a microseismic signal, the microseismic signal is filtered. The algorithm is simple and easy and an effect is ideal. Effective noise reduction filtering can be performed on a mine noisy microseismic signal. And a good technical value and an application prospect are possessed.

Function spaces defined by scaling functions are used to generate bandlimited noise octaves and other attribute data sets. Scaling functions are basis functions that admit multiresolution analysis and include piecewise constant scaling functions, piecewise polynomial scaling functions, bandlimited scaling functions, Daubeschies scaling functions, as well as other multiresolution analysis scaling basis functions known to those of skill in the art. Scaling basis functions can be locally supported or have infinite support. The properties of the scaling basis functions used to construct bandlimited noise octaves may ensure that any bandlimited noise octave at resolution level N is orthogonal to bandlimited noise octaves and their associated scaling basis functions at all resolution levels less than N. Bandlimited noise octaves can be scaled to any resolution level and guaranteed to have no effect on images at any lower resolution level.

A neural network system for predicting time series data. The system receives analyzed data obtained by multiresolution analysis of the time series data. The input processing layer of the system includes a series of neurons corresponding to the different levels of analysis, each neuron receiving the analyzed data for it own level. The output of each of these neurons is supplied as an additional input to the neuron for the next lower level of analysis. A predicted value is derived from the output of the neuron at the lowest level. The passing of results from one level to another improves prediction accuracy and simplifies the structure of any further processing layers.

An electrophotographic photoconductor having a photosensitive layer and a crosslinked resin surface layer over a support, wherein shapes of concaves and convexes in a surface of the electrophotographic photoconductor are measured by a surface roughness / profile measuring device to obtain one-dimensional data arrays, the arrays are subjected to multiresolution analysis (MRA-1) through wavelet transformation to be separated into six frequency components including HHH, HHL, HMH, HML, HLH and HLL to obtain one-dimensional data arrays, the arrays of the HHL are thinned out to be reduced 1 / 10 to 1 / 100, thereby producing one-dimensional data arrays, which are then subjected to multiresolution analysis (MRA-2) through wavelet transformation to be separated into six frequency components including LHH, LHL, LMH, LML, LLH and LLL to thereby obtain 12 frequency components in total; and a center-line average roughness (WRa) of the 12 frequency components satisfies relationship (i) below.1−597×WRa(HML)+238×WRa(HLH)−95×WRa(LHL)+84×WRa(LMH)−79×WRa(LML)+55×WRa(LLH)−17×WRa(LLL)>0  (i)

The invention relates to an industrial device rapid locating method based on machine vision, which is a device locating method adopted by machine vision software running in a hardware environment built by an industrial lens, an industrial camera, an image gathering card and an industrial personal computer. The method provided by the invention is a method with combined multiresolution analysis and wavelet transformation characteristics, and the method comprises the following steps of: firstly locating approximately according to the projection transformation vectors of horizontal and vertical direction coefficients of wavelet transformation, then implementing template matching in a low-resolution image, and finally accurately locating in the approximately located field of the original image. The application effect of the method in semiconductor and electronic manufacturing industry is excellent, the locating speed is particularly and obviously improved by means of multiresolution analysis and location preparation, and the locating process of a wafer can be completed in about 10 milliseconds.

The invention discloses an SAR image change detection method based on NSCT domain synthetic kernels to mainly solve the problem that SAR image change detection can not be conducted through the multiresolution analysis due to existing Gabor domain synthetic kernels. The SAR image change detection method includes the steps of firstly, conducting NSCT decomposition in multiple dimensions and multiple directions on original two-time-phase SAR images; secondly, extracting the normalized strength characteristic of the decomposed two-time-phase images in each dimension and the normalized texture characteristic of the decomposed two-time-phase images in each dimension, and constructing the strength differential value synthetic kernel and the texture differential value synthetic kernel in each dimension; thirdly, enabling the differential value synthetic kernels in any dimension to be inputted into a support vector machine to be detected, and obtaining a change detection result of the dimension; fourthly, conducting determination-level inter-dimension fusion on the change detection results of all the dimensions, and obtaining a final change detection result. Compared with a method based on the existing Gabor domain synthetic kernels, the SAR image change detection method based on the NSCT domain synthetic kernels is high in detection accuracy and high in calculation efficiency and can be used for SAR image change detection.

The invention discloses a synthetic aperture sonar speckle noise suppression method based on multiresolution analysis. The method comprises the steps that 1, speckle noise in an original synthetic aperture sonar image is converted into additive noise through logarithm processing, and a logarithm processing image is obtained; 2, wavelet threshold de-noising processing is conducted on the logarithm processing image, and a reconstituted image is obtained; 3, index processing is conducted on the reconstituted image. The synthetic aperture sonar speckle noise suppression method based on multiresolution analysis solves the problem that synthetic aperture sonar speckle noise in a synthetic aperture sonar image cannot be processed in the prior art. The synthetic aperture sonar speckle noise suppression method based on multiresolution analysis improves the gray scale contrast ratio of the target and background of the synthetic aperture sonar image with the target, a subsequent image processing process is facilitated, the edges of different substrate areas of the landform synthetic aperture sonar image are made to be more obvious, and therefore the subsequent image processing such as image segmentation and substrate characteristic analysis is facilitated.

The invention discloses an extraction method and equipment of apparent aluminum oxide concentration for process control of an aluminum electrolysis cell. The method comprises the following steps: A, separating normalized bath voltage data with any sample frequency layer by layer to obtain signal components relevant to an aluminum oxide concentration state by means of a multiresolution analysis theory and shift-invariant wavelet transform; B, defining a state characteristic variable relevant to the apparent aluminum oxide concentration state information as the apparent aluminum oxide concentration according to the physical meaning of the signal components relevant to the aluminum oxide concentration state; and C, designing a novel slope by means of the obtained apparent aluminum oxide concentration. Experiments verify that a rate of change of the apparent aluminum oxide concentration is obtained, influence of changing factors, for example, electrolyte components, electrolyzing temperature, polar distance and molten aluminum fluctuation, which cannot be detected in real time on pre-estimation of the aluminum oxide concentration in industrial production can be eliminated greatly, andrelative change of the aluminum oxide concentration is represented relatively well.

A method for discriminating between explosive events having their origins in High Explosive or Chemical / Biological detonation employing multiresolution analysis provided by a discrete wavelet transform. Original signatures of explosive events are broken down into subband components thereby removing higher frequency noise features and creating two sets of coefficients at varying levels of decomposition. These coefficients are obtained each time the signal is passed through a lowpass and highpass filter bank whose impulse response is derived from Daubechies db5 wavelet. Distinct features are obtained through the process of isolating the details of the high oscillatory components of the signature. The ratio of energy contained within the details at varying levels of decomposition is sufficient to discriminate between explosive events such as High Explosive and Chemical / Biological.

An adjustable mixed classification method of spectrum and space character includes setting a series of multimeasure windows of central pixel and using Mallat transform to set up multiresolution analysis of each window, obtaining four wavelet subbands by each time of Mallat decomposition on each window, using four said subbands to structure a rotary unchanged space character, selecting sample similar to said type and using mean value and variance measure to structure deflection vector according to spectrum and space information of sample, making character weight on different space character and selecting occupied proportion of spectrum and space character in classification.

The invention discloses a network community detection method based on an M elite coevolution strategy, wherein the network community detection method solves the problems that in the prior art, the convergence rate is low and easily lapses into the local optimum, multiresolution analysis of a network structure cannot be achieved. The implementation steps include that (1) network data are loaded; (2) network community populations are initialized; (3) the network community populations are divided; (4) a network community team is organized; (5) candidate network community division is detected; (6) the network community populations are updated; (7) local network communities are detected; (8) the network community populations are updated; (9) whether iteration is terminated or not is judged; (10) a network community detection result is output. When the network community detection method is used for detecting community structures in a network, expanded module density functions serve as fitness functions, a network structure is analyzed with different resolutions, and the convergence rate is quickened through leading-in of local detection and does not easily lapse into the local optimum.

An electrophotographic photoreceptor including a conductive substrate, a photosensitive layer, and a surface layer, which satisfies the following inequations: 0.005<WRa(LMH)<0.03  (i) 0.010<WRa(LHH)<0.03  (ii) 0.005<WRa(LML)<0.20  (iii) WRa(LLH)>WRa(LMH)  (iv) WRa(LLH)>WRa(LHH)  (v) wherein WRa (μm) represents a center-line average roughness of frequency components LHH, LHL, LMH, LML, LLH, and LLL that are obtained by subjecting a one-dimensional data array of a surface profile of the electrophotographic photoreceptor to wavelet transformation multiresolution analysis so as to be separated into 6 frequency components; thinning a one-dimensional data array of the lowest frequency component so that the number of data array is reduced to 1 / 40; and subjecting the thinned one-dimensional data array to wavelet transformation multiresolution analysis so as to be separated into the 6 frequency components LHH, LHL, LMH, LML, LLH, and LLL.

The invention provides an iris image matching method which is based on a multi-resolution analysis. The method comprises the following steps that: firstly an original normalized iris image is extracted through positioning and normalizing; then illumination estimation and histogram enhancement are carried out to the original normalized iris image; next, the enhanced normalized iris image is divided into 8 horizontal tapes, and the approximation A2-2dij(x) of the gradation sequence dij(x) of each tap intermediate row under the resolution ratio of 2<-2> is extracted to compose a characteristic vector; when two iris images are matched, the Euclidean distance between the characteristic vectors of the two images is calculated and is compared with the preset threshold value, thereby can judge whether the two iris images belongs to the same person. Adopting the iris image matching method of the invention can have stabler performance, and can avoid the influence of the high-frequency noise.

The invention discloses a micro-grid harmonic wave detection method. The micro-grid harmonic wave detection method includes the following steps: 1) a micro-grid harmonic wave signal acquisition systemacquires a voltage and current harmonic wave signal of a common point of micro-power source connection; 2) the acquired voltage and current harmonic wave signal is decomposed and reconstructed according to an idea of multiresolution analysis in wavelet transformation by using a tower method in the wavelet transformation to acquire a series of high frequency signals and low frequency sub-signals;3) the acquired low frequency sub-signals are decomposed into a series of intrinsic mode function components by using an empirical mode decomposition method, and fundamental components and high-orderharmonic components are extracted from the intrinsic mode function components; and 4) Hilbert-Huang transform is performed on the extracted fundamental components and high-order harmonic components toacquire information of amplitudes, frequencies and phases of harmonic waves. The micro-grid harmonic wave detection method overcomes the problem of dependency of the wavelet transformation on selection of basic wavelets and improves frequency spectrum aliasing of the wavelet transformation, thus harmonic waves of currents in micro-power source connection can be detected in a more real-time and accurate manner.

A signal detection device allows wavelet transformation of an object signal to be performed in real time by using a real signal mother wavelet. The signal detection device has: an object signal decomposition unit having a lifting scheme structure or a multiple analysis structure relying on multiresolution analysis; a parasitic filter coupled to a desired decomposition filter of the object signal decomposition unit, with the parasitic filter being configured such that a real signal mother wavelet is inputted to the object signal decomposition unit and a generic discrete wavelet transformation is performed, the parasitic filter substantially reproduces and outputs the inputted real signal mother wavelet, and with the real signal mother wavelet being made up of the object signal; means for inputting the object signal to the object signal decomposition unit and performing discrete wavelet transformation using the real signal mother wavelet; and means for computing a wavelet instantaneous correlation on the basis of an output of the parasitic filter.

The invention provides an electric energy quality regulation method, voltage temporary change detection method and dynamic voltage restorer. The voltage temporary change detection method comprises thefollowing steps: acquiring an electric grid voltage signal, selecting a decomposition scale, performing initial wavelet transformation, and acquiring wavelet coefficients of a first set frequency band and a second frequency band; performing single reconstruction for the wavelet coefficient of the first frequency band, and acquiring a fundamental wave signal of the electric grid voltage; performing secondary wavelet transformation, acquiring a first frequency band component and a second frequency band component; performing single reconstruction for the second frequency band component, and determining start and stop time of occurrence of voltage temporary change; and performing single reconstruction for the first frequency band component, and determining fundamental wave electric grid voltage temporary amplitude information. Double wavelet transformations are used for detection of the voltage signal containing temporary change and harmonic waves, based on wavelet transformation multiresolution analysis characteristics and singularity detection property, decoupling of harmonic waves and temporary change electric energy quality problems is solved, noise influence is restrained, accurate positioning of a temporary change signal is completed, time delay can be reduced effectively, and influence on electric grid real-time compensation is reduced.

Function spaces defined by scaling functions are used to generate bandlimited noise octaves and other attribute data sets. Scaling functions are basis functions that admit multiresolution analysis and include piecewise constant scaling functions, piecewise polynomial scaling functions, bandlimited scaling functions, Daubeschies scaling functions, as well as other multiresolution analysis scaling basis functions known to those of skill in the art. Scaling basis functions can be locally supported or have infinite support. The properties of the scaling basis functions used to construct bandlimited noise octaves may ensure that any bandlimited noise octave at resolution level N is orthogonal to bandlimited noise octaves and their associated scaling basis functions at all resolution levels less than N. Bandlimited noise octaves can be scaled to any resolution level and guaranteed to have no effect on images at any lower resolution level.

The invention relates to a heart head and neck blood vessel combined imaging method, comprising the steps of: through a module coupling mode, determining noncollinear coupling characteristic points on a heart head and neck blood vessel image to be tested and a standard image; separately establishing the three-dimensional coordinate system of the standard image and the three-dimensional coordinate system of the heart head and neck blood vessel image to be tested based on the characteristic points; separately performing multiresolution analysis on the standard image and the heart head and neck blood vessel image to be tested in the three-dimensional coordinate systems to obtain multiresolution sub-images; starting registering from a sub-image of lowest resolution, and performing registering by degrees until obtaining registration parameters of the image of highest resolution and the heart head and neck blood vessel image to be tested after registering; and performing Force double source CT coarse pitch scanning on the heart head and neck blood vessel image to be tested, and fusing with the heart head and neck blood vessel image to be tested after registering to obtain the combined imaging of the heart head and neck blood vessel image to be tested. The heart head and neck blood vessel combined imaging method obtains the combined imaging of the heart head and neck blood vessel image to be tested through different modes.

The image-processing unit (16) of an embodiment is provided with an expansion unit (161), a decomposition unit (162), a data-processing unit (163), a reconstruction unit (164), and a contraction unit (165). The expansion unit (161) expands the data to be processed with a magnification that is based on the optimal sample number, which is the ideal number of samples per unit length for image-processing. The decomposition unit (162) decomposes the expanded data using multiresolution analysis to generate data sets with the optimal number of samples. The data-processing unit (163) performs image processing on the data sets with the optimal number of samples to generate processed data sets. The reconstruction unit (164) reconstructs the processed data sets using multiresolution analysis to generate reconstructed data. The contraction unit (165) contracts the reconstructed data so that the number of samples per unit length equals the number of samples per unit length of the data that was processed.

The invention discloses a vein recognition method based on texture features. The method includes the steps of firstly, creating a hand vein image library, images in each category being randomly divided into a training set and a testing set, and the number of images in the training set being the same as the number of images in the testing set; secondly, acquiring and pre-processing hand vein images; and thirdly, recognizing the vein images based on the texture features. The method of the invention is capable of targeted extraction based on the characteristics of vein directivity, has strong robustness and rotation invariance, and can achieve multi-resolution and multi-directional analysis. Meanwhile, the problem of feature dimension is solved and excellent effects are gained in an actual test.

The present invention is an online methodology for end point detection for use in a chemical mechanical planarization process which is both robust and inexpensive while overcoming some of the drawbacks of the existing end point detection approaches currently known in the art. The present invention provides a system and method for identifying a significant event in a chemical mechanical planarization process including the steps of decomposing coefficient of friction data acquired from a chemical mechanical planarization process using wavelet-based multiresolution analysis, and applying a sequential probability ratio test for variance on the decomposed data to identify a significant event in the chemical mechanical planarization process.

A signal processing unit for playing back an audio signal at a decelerated rate keeping pitch constant. The audio signal is at least one of a speech signal, a pure music or an audio signal which comprises of both speech and music signal. The signal processing unit comprises a plurality of bandpass filters with each of them receiving a first plurality of samples of the audio signal, a plurality of interpolators and an adder. The plurality of bandpass filters generate a second set of plurality of samples after passing the first plurality of samples of the audio signal through each of them. The plurality of bandpass filters have different pass bands, different stop bands, and a constant Q factor. The plurality of interpolators are connected to the plurality of bandpass filters and generate a third set of plurality of samples. The plurality of bandpass filters and the plurality of interpolators correspond in number. The adder superimposes constituents of the third set of plurality of samples generated by the plurality of interpolators. The adder outputs a fourth plurality of samples which on playing gives rise to a decelerated version of the audio signal having a pitch which is consistent with a pitch of the audio signal in a non-decelerated condition.

The invention relates to an industrial device rapid locating method based on machine vision, which is a device locating method adopted by machine vision software running in a hardware environment built by an industrial lens, an industrial camera, an image gathering card and an industrial personal computer. The method provided by the invention is a method with combined multiresolution analysis andwavelet transformation characteristics, and the method comprises the following steps of: firstly locating approximately according to the projection transformation vectors of horizontal and vertical direction coefficients of wavelet transformation, then implementing template matching in a low-resolution image, and finally accurately locating in the approximately located field of the original image. The application effect of the method in semiconductor and electronic manufacturing industry is excellent, the locating speed is particularly and obviously improved by means of multiresolution analysis and location preparation, and the locating process of a wafer can be completed in about 10 milliseconds.