47 results about "Stationary wavelet transform" patented technology

The invention discloses an SAR (Synthetic Aperture Radar) image segmentation method based on multi-scale feature fusion, comprising the steps of: firstly, extracting the textural features of an image by utilizing fast discrete curvelet transform (FDCT), extracting the statistic features of the image by utilizing stationary wavelet transform (SWT), then, fusing the two type of multi-scale features into high-dimensional feature vectors, and finally, segmenting by adopting a fuzzy C-mean clustering method. At the same time of eliminating broken pieces in a homogenizing area, the boundary is more precise and smoother. The method is favorable for effectively preventing the extraction of textural information from the interference of speckle noises. By fusing SWT coefficient statistical features, the textural information and the grey statistical information of the SAR image are effectively utilized and a favorable segmentation effect can be achieved.

The invention discloses a natural image denoising method based on the regional division, which mainly solves the problem that patches exist in an image after an image is donoised through the existing three-dimensional block matching denoising algorithm. The method comprises the following steps: firstly, performing two-dimensional stationary wavelet transform on an input image to be denoised, and performing inverse transform on a high frequency coefficient subjected to zero setting, so as to obtain a reconstructed image; secondly, exacting the structural information of the reconstructed image, so as to obtain an image structure sketch; thirdly, dividing the noisy original image into a structure region, a smooth region and a non-smooth region through statistic features of an image block and the image structure sketch; fourthly, performing denoising on the smooth region with an improved non-local mean method, performing denoising on the non-smooth region with a BM3D (Block matching 3D) method, and performing denoising on the structure region with a directional feature-based BM3D method; and fifthly, combining estimated results of the structure region, the smooth region and the non-smooth region, so as to obtain the final denoised image. The method can be used for preprocessing natural images.

The invention discloses a method for extracting the feature point of an electrocardiosignal waveform. The method comprises the steps of determining positions of a QRS wave point, a P wave point and a T wave point, wherein the position of the QRS wave point is determined by processing an electrocardiosignal by stationary wavelet transform, determining the QRS wave represented optimal target layer and T and P wave represented optimal target layers, finding out a maximum value and minimum value pair of the corresponding target layer, removing unqualified maximum value and minimum value pairs, carrying out error inspection and miss inspection on the position of the R wave point to obtain the final position of the R wave point, determining positions of the Q wave point and the S wave point, and determining positions of the P wave point and the T wave point according to the QRS wave determined by the T and P wave represented optimal target layers. According to the method, stationary wavelet transform is used in position determination of the QRS wave point, the T wave point and the P wave point, and can be used for effectively avoiding resolution damage when the dimension is increased in comparison with discrete wavelet transform, movement is not deformed, and the problem in the prior art can be effectively solved.

The invention discloses an image super-resolution reconstruction method based on wavelet coefficient learning, belonging to the field of image super-resolution. the method includes selecting that image train set and the test set; obtaining The wavelet coefficients of LR image and HR image by two-dimensional stationary wavelet transform of low-resolution image and two-dimensional wavelet packet transform of high-resolution image respectively. 3, constructing a depth neural network; 4, extracting that depth network feature of the wavelet coefficients of the LR image by use the depth neural network, and obtaining the wavelet residual coefficients of the HR image; Step 5, restoring the obtained wavelet residual coefficients of the HR image to the wavelet coefficients of the HR, and carrying out the two-dimensional wavelet packet inverse transform to obtain the final high-resolution image. Compared with the prior art, the invention can perform super-resolution reconstruction on an image ina very short time by training a depth super-resolution model by using a large number of external training sets, and the reconstruction result is obviously superior to most algorithms.

The invention discloses a fetal electrocardiogram signal extracting method based on wavelet threshold denoising. The method comprises the following steps of: carrying out stationary wavelet transform processing on a maternal abdomen signal, and decomposing to obtain each layer of wavelet coefficient; independently processing each layer of wavelet detail coefficient, removing fetal electrocardiogram wavelet coefficients, reconstructing remained maternal electrocardiogram wavelet coefficients, and obtaining maternal electrocardiogram in the abdomen signal; removing the reconstructed maternal electrocardiogram from the abdomen signal to obtain a fetal electrocardiogram signal; and denoising the extracted fetal electrocardiogram by utilizing a wavelet correlation denoising algorithm to obtain a clear fetal electrocardiogram signal. Due to the method, the accurate extraction of the fetal electrocardiogram signal can be realized. The method is based on a single channel, the computation of wavelet transform modulus maxima is avoided, and better instantaneity is realized.

The invention discloses a stationary wavelet transform denoising method based on cross validation. The stationary wavelet transform denoising method comprises the following steps: determining an optimal threshold value in combination with cross validation and stationary wavelet transform at first, and then, denoising a spectrum by utilizing the optimal threshold value on the basis of the stationary wavelet transform. By means of the stationary wavelet transform denoising method, a comparatively accurate threshold value can be obtained; noise can be removed effectively; the Gibbs phenomenon can be avoided; and spectral peak features cannot be influenced.

The invention provides a stall inception identification method of an axial-flow compressor. Firstly, the pulsating pressure signal of the axial-flow compressor is subjected to band selection and filtering through a stationary wavelet transform; and secondly, the correlation between the filtered pulsating pressure signal at a rotor period and the signal at a previous rotor period is compared, and a cross correlation coefficient is calculated. Through monitoring the change of the cross correlation coefficient, a stall inception characteristic can be effectively identified and monitored, and the aim of stall early warning is achieved.

The invention discloses a foggy day image salient target detection method, which comprises the following steps of: step 1, performing color space conversion on a foggy day image in a frequency domainto calculate the saliency of the foggy day image, and solving a frequency domain saliency map; step 2, performing super-pixel segmentation on the foggy day image in a spatial domain, calculating the significance of each super-pixel block, and solving a spatial domain saliency map; step 3, fusing the saliency map of the image in the frequency domain and the saliency map of the image in the space domain into a saliency map through discrete stationary wavelet transform; step 4, obtaining a contour map of the foggy day image through the target contour detection model; and step 5, adding the saliency map fused based on the frequency domain and the spatial domain to the contour map to obtain a final saliency map. According to the method, a traditional machine method and a deep learning method are combined, the robustness of traditional significant target detection is improved, and a significant target in a foggy day scene can be efficiently and accurately detected; meanwhile, for some imageswith complex backgrounds, a significance target can be well extracted.

The invention discloses a laser radar real-time imaging and building characteristic extracting method comprising the following steps of: mapping laser pin spots of a scanning line obtained by a laser radar to an ideal scanning line; carrying out rasterization on the ideal scanning line; obtaining elevation information at each grid position on the ideal scanning line through a linear interpolation method; carrying out graying treatment on the elevation information in the ideal scanning line, so that a probing area corresponding to the scanning line is imaged; and extracting building characteristic information in the ideal scanning line by using discrete stationary wavelet transform. By utilizing the method provided by the invention, the scanning lines of the laser pin spots in point cloud data of the laser radar can be processed one by one, so that the operations of real-time imaging, building characteristic extraction and the like in a scanning process of the laser radar can be realized.

The invention relates to an optical image denoising method based on serial connection of curvelet transform and wavelet transform. The method comprises the steps of 1, selecting a wave band, with noise, of optical data as an image to be processed; 2, carrying out forward Curvelet transform on the image to be processed to obtain multi-scale and multi-direction Curvelet transform coefficients; step 3, with noisy data in the Curvelet transform coefficients to be removed through wavelet transform, carrying out wavelet transform denoising and then carrying out reverse wavelet transform to obtain denoised Curvelet coefficients; 4, carrying out reverse Curvelet transform on the denoised Curvelet coefficients to obtain a denoised reconstructed image. Compared with single curvelet transform or single wavelet transform, the image denoised through the method has a higher peak signal-to-noise ratio and a better visual effect, the stripe noise of the optical image is removed, detail information of the image is well reserved, and the quality of the image is improved.

The invention discloses an SAR and optical image fusion method for landslide detection. The method provided by the invention comprises the following steps: step 1, preprocessing an SAR and an opticalimage; step 2, performing HIS conversion on the optical image to obtain three components I, H and S; step 3, performing stationary wavelet transform and high-frequency component energy maximization fusion on I components of the SAR image and the optical image; step 4, performing saliency detection on the low-frequency and high-frequency components of the SAR image and the gray scale information ofthe image, establishing an SAR salient target detection area guidance function, and partitioning the SAR image; step 5, establishing a salient region fusion rule, and realizing image fusion accordingto a regional fusion strategy; and step 6, identifying and extracting landslide disaster information based on the fused image. The method has good adaptability to SAR and optical image fusion for landslide detection, adopts related processing measures in structure maintenance, noise removal and spectrum reservation, and obtains an excellent effect.

The invention discloses a high-voltage direct-current transmission line calculation model based on distributed resistance parameters. The method comprises the steps of (1) acquiring a fault voltage signal and a current signal; (2) carrying out decoupling to obtain aerial mode components; (3) calculating voltage and current aerial mode components of a direct-current transmission line along the lineby utilizing a distributed parameter line model based on distributed resistance; (4) calculating a reverse traveling wave current of the direct-current transmission line along the line; and (5) calculating a modulus maximum value of the reverse traveling wave current by adopting stationary wavelet transform. According to the method, after voltage and current values of measuring points at the twoends of the line are extracted, voltage and current aerial mode component values of the direct-current transmission line along the line are calculated by utilizing the line model based on the distributed resistance; then according to the characteristics that a first reverse traveling wave is not reflected yet, the influence of frequency change of a reflection coefficient is avoided and fault features are obvious, a reverse traveling wave current value of the direct-current transmission line along the line is calculated; and finally, the modulus maximum value of the reverse traveling wave current is calculated through the stationary wavelet transform, so that the accuracy of the transmission line calculation model is improved.

The invention discloses a stationary wavelet transform (SWT) improved particle filter-based synthetic aperture radar (SAR) image despeckling method, which mainly solves the problems of low statistical modeling accuracy, fuzzy image edge and texture after despeckling and the like in a conventional discrete wavelet transform (DWT) domain basic particle filter method. The realization process of the method comprises the following steps of: (1) transforming an airspace image to be despeckled to obtain a stationary wavelet domain and extracting SWT domain image groups with different transform scales in horizontal, vertical and diagonal directions; (2) performing despeckling processing on the SWT domain image groups by using an improved particle filter despeckling method; and (3) transforming the despeckled stationary wavelet domain image groups to airspace by using stationary wavelet inverse transform, wherein the airspace image is a finally despeckled result. Compared with the conventionalDWT domain basic particle filter method, the SWT improved particle filter-based SAR image despeckling method has a stable result, an obvious despeckling effect and remarkable image texture information, single point target and boundary maintaining effects, and can be used for target detection and target identification.

The invention discloses a direct-current transmission line protection method based on initial voltage traveling wave frequency domain attenuation rate. The method is characterized by comprising the following steps of: measuring voltages at head end protection installation positions of a positive electrode and a negative electrode of a direct-current line, and constructing a low-voltage starting criterion; if protection is started, measuring the currents at the head end protection installation positions of the positive electrode and the negative electrode of the direct-current line, and identifying a fault direction according to the symbols of the current fault component integral value of the positive electrode or the negative electrode; if the fault is a forward fault, calculating the line-mode voltage traveling wave of the direct current line, performing stationary wavelet transform of different scales on the line-mode voltage traveling wave, and extracting a wavelet transform modulusmaximum value under the corresponding scale; and calculating the equivalent attenuation rate in the initial voltage traveling wave frequency domain, and identifying the faults inside and outside theline area according to the absolute value of the rate. According to the method, the requirement for rapid main protection of the direct-current line is met, and the internal high-resistance fault andthe external fault can be reliably recognized with high sensitivity.

The invention discloses an image enhancement method and system, and the method comprises the steps: carrying out the super-resolution of an input image I, obtaining an image I ', carrying out the stationary wavelet transformation of the image I', and obtaining a low-pass sub-band L; carrying out stationary wavelet transform on the input image I to obtain detail sub-bands Hh, Hv and Hd, then respectively enhancing the detail sub-bands Hh, Hv and Hd, and then carrying out super-resolution; and performing inverse stationary wavelet transform on the low-pass sub-band L and the detail sub-bands Hh,Hv and Hd after super-resolution to obtain an enhanced image. The system comprises a first super-resolution module, a first stationary wavelet transform module, a second stationary wavelet transformmodule, a detail sub-band enhancement module, a second super-resolution module and an inverse stationary wavelet transform module. According to the image enhancement method and system based on the improved FSRCNN, the obtained enhanced image is clear in detail, high in contrast ratio, good in visual effect, excellent in objective evaluation index and good in real-time performance.

The invention discloses an OTDR curve data analysis method based on wavelet transform dynamic noise reduction, and belongs to the data analysis technology field. The method includes: performing discrete stationary wavelet transform on OTDR data by adopting a Haar wavelet basis; Extracting all undetermined events, calculating a line average consumption threshold; using a least square method and anevent attenuation threshold, fiber breakage threshold, end height threshold parameters, to accurately position the event in the undetermined event, determining whether to perform noise reduction and determine a section needing to be subjected to noise reduction, performing section noise reduction processing by adopting a wavelet basis DB3 modulus maximum value method, and analyzing the section data subjected to noise reduction by adopting wavelet transformation and a least square method again. The OTDR curve data can be effectively analyzed by combining wavelet transformation and the least square method, a large number of optical cables are subjected to little noise interference in practical application, the noise interference is also applied to some sections of the optical cables, and theOTDR curve data analysis speed is greatly increased.

The invention discloses an improved wavelet transform and convolutional neural network image denoising method. The method comprises the following steps of 1, performing size normalization, noise addition and data expansion on an original digital image; 2, constructing an image denoising network model WTCNN, constructing a wavelet coefficient model in order to obtain features of a noise-containing image, transforming the input noise-containing image into wavelet coefficients of different sub-bands by adopting stationary wavelet transform SWT, obtaining input data of a convolutional neural network, and then constructing a model structure combining two image denoising algorithms; and 3, performing simulation denoising on a noise-containing image in a real scene through the ideal denoising model obtained in the step 2, and obtaining a denoised image. According to the invention, the network performance of the denoising algorithm is improved.

The embodiment of the invention discloses a robust combined domain color image zero-watermark embedding and extracting method, which comprises the following steps of: firstly, carrying out two-dimensional stationary wavelet transform on a brightness component of an original color image in a YIQ color space, and carrying out non-overlapping partitioning on a transformed low-frequency sub-image; and then selecting a certain number of image blocks according to the size of the watermark to carry out all-phase discrete cosine biorthogonal transformation, and finally constructing robust features by using coefficient symbols after block transformation so as to realize zero-watermark embedding. A watermark extraction method is similar to a watermark embedding method, but a watermark needs to carry out rotation discrimination and correction on a to-be-authenticated image in advance. The method has good robustness for resisting various common image processing attacks, such as noise addition, filtering, JPEG compression, zooming and the like, and rotation attacks at any angle.

The invention discloses a speckle interference image noise reduction method and device. The method comprises the steps of converting a discontinuous original phase diagram into a continuous sine phasediagram and a cosine phase diagram through sine transformation and cosine transformation; performing noise reduction processing on the sine phase diagram and the cosine phase diagram by using stationary wavelet transform to obtain a sine noise reduction phase diagram and a cosine noise reduction phase diagram; and determining a target noise reduction phase diagram according to the sine noise reduction phase diagram and the cosine noise reduction phase diagram. According to the invention, the noise in the speckle interference image can be effectively and quickly filtered.

The invention discloses a multi-modal medical image fusion method, which comprises the following steps of 1) source image decomposition: performing two-stage decomposition discrete stationary wavelet transform on at least two source images, 2) source image fusion: fusing the at least two source images through seven enhanced radial basis function neural networks, and 3) inverse wavelet transform: performing inverse wavelet transform, through the inverse wavelet transform, converting the seven fusion sub-bands into a fusion image. The method has the advantages that through translation of the invariant multi-scale transformation operator, enough direction information is provided, and the algorithm is relatively simple and easy to operate. In order to enhance the self-learning ability of the neural network, in combination with medical image fusion, pixel values, regional energy, pixel gradients and regional average gradients are adopted to form an input layer of the neural network, information of target feature points is accurately extracted, and meanwhile, information loss is avoided.

The invention relates to a method for detecting P and T characteristic waves of an electrocardiosignal. The method comprises the following steps of: a, acquiring the electrocardiosignal of a human body to perform pre-processing; b, carrying out smoothing processing on the signals under the scale of the to-be-detected sections of the P and T characteristic waves through stationary wavelet transform; c, carrying out electrocardio data P and T characteristic wave peak value point detection by utilizing a maximum and minimum value principle; d, correcting the positions of P and T characteristic wave peak points of the electrocardio data according to a time shift correction principle; and e, evaluating a detection result by calculating three indexes of the error rate e<r>, the sensitivity S<e>and the correct prediction degree P of P and T characteristic wave detection. The electrocardiosignal is processed and smoothed by the method in the invention; therefore, the noise is effectively filtered; furthermore, low-frequency characteristic waves of the electrocardiosignal are recovered; and effective information of the electrocardiosignal is reserved.

The invention discloses an identification method of hemispherical and conical models based on discrete stationary wavelet transform. The method includes: firstly, obtaining point cloud data of a measured curved surface by employing a three-dimensional scanner; then obtaining an elevation image of the measured curved surface and elevation sequences of pixels of various rows and columns in the image by employing a gridding method; processing the obtained elevation sequences by employing discrete stationary wavelet transform to obtain corresponding wavelet detail coefficient sequences; further deducing an ideal mathematical expression of the measured curved surface according to corresponding relations between the elevation sequences and the wavelet detail coefficient sequences; calculating ideal elevation sequences of the measured curved surface and corresponding wavelet detail coefficients according to the estimated ideal mathematical expression; and finally analyzing wavelet detail coefficient sequences corresponding to the elevation sequences of actual measurement data and the wavelet detail coefficient sequences corresponding to the ideal mathematical expression of the measured curved surface to realize identification of a measured model.

The invention provides a wavelet dual-threshold denoising method based on an interlayer correlation coefficient, which comprises the following steps of: processing a noisy signal by adopting stationary wavelet transform (SWT) to obtain an output coefficient of a low-frequency part and an output coefficient of a high-frequency part, and according to each group of lower threshold regulation factors represented by each firefly, processing the output coefficient of the high-frequency part by adopting a wavelet dual-threshold function based on an interlayer correlation coefficient, then reconstructing the processed output coefficient of the high-frequency part and the processed output coefficient of the low-frequency part to obtain a denoised signal, and finally taking the signal-to-noise ratio of the denoised signal as the brightness of fireflies. The higher the signal-to-noise ratio is, the higher the brightness of the fireflies is, the better the denoising effect of the lower threshold adjustment factor represented by each firefly is, and the optimal denoising signal is obtained by using the firefly algorithm. According to the method, a layered threshold value and a semi-soft threshold value method are combined, the firefly algorithm is added, and given threshold values are optimized through a signal-to-noise ratio (SNR) index; and the threshold value precision is improved.

A micro-fluidic chip signal denoising method based on an improved threshold function comprises the following steps: (1) selecting a wavelet basis, determining the number of decomposition layers, and carrying out discrete stationary wavelet transform on a noisy micro-fluidic chip signal to obtain a wavelet coefficient; (2) selecting a proper threshold value, and performing threshold value quantization processing on the wavelet coefficient by adopting an improved threshold value function; and (3) carrying out discrete stationary wavelet inverse transformation on the wavelet coefficient subjectedto threshold quantization processing to obtain a denoised micro-fluidic chip signal. The problems that a reconstruction signal is oscillated due to discontinuity of a hard threshold method and the reconstruction signal is distorted due to constant deviation of a soft threshold method are solved. The signal denoising effect of the micro-fluidic chip can be improved, the denoised signal baseline ofthe micro-fluidic chip is smooth, the characteristics are well reserved, and the requirement of quantitative analysis of the signal of the micro-fluidic chip is met.

The invention relates to a pipeline anti-corrosion layer tiny damage point positioning method based on stationary wavelet transform. The method comprises the following steps of S1, obtaining to-be-detected pipeline power-off potential signals of a plurality of point positions; S2, performing stationary wavelet transform processing on the obtained power-off potential signals to obtain a detail coefficient containing damage information; S3, performing soft threshold de-noising processing on the detail coefficient containing the damage information to obtain a new coefficient; S4, obtaining information of a pipeline anti-corrosion layer damaged point through the coefficient, marking the point, and re-planning a measurement interval by taking the point as a center; and S5, cycling steps S2-S4, and determining the position of the anti-corrosion layer damaged point. According to the method, the information of the pipeline anti-corrosion layer tiny damage point can be effectively represented, and when the distance exists between the damage point and the measurement point of a reference electrode, a damage point signal near the measurement point can also be identified.