66results about How to "Improve denoising efficiency" patented technology

The invention discloses a novel K value optimization method in a point cloud clustering denoising process. The method comprises the steps that (1) a three-dimensional laser scanning instrument obtains space sampling points of the surface of an actual object; (2) the space sampling points are used as K values to optimize a clustering sample for clustering, a K-means clustering method is used for generating different clustering results of point cloud clustering in a clustering number search range, clustering validity indexes are used for evaluating different clustering results, and an obtained best clustering number is used as the optimal K value; (3) the optimal K value is used as a clustering initial value of three-dimensional point cloud clustering denoising, and three-dimensional point clouds are subjected to clustering; and (4) local outlier noise points are identified and removed by carrying out Euclidean-distance-based threshold value judgment in a class of clustering results, and ideal point clouds are obtained. The novel K value optimization method is used, the value is used for carrying out optimization clustering on point clouds with noise, so that the denoising accuracy of ideal point clouds is high, denoising speed is increased, and a later-period reconstructed three-dimensional model is smooth and real.

The invention discloses an image denoising method and system based on deep learning. The image denoising method based on deep learning includes the steps: constructing a main neural network structureand an auxiliary neural network structure, respectively assigning the trainable parameter initial value of the first convolutional layer and the trainable parameter initial value of the fifth convolutional layer in the auxiliary neural network structure to the trainable parameter initial value of the first convolutional layer and the trainable parameter initial value of the 15th convolutional layer in the main neural network structure; adding a training set noise adding image into the main neural network structure after assignment, and obtaining a noise characteristic image by performing imagecharacteristic extraction, training and learning on the input training set noise adding image through a forward propagation algorithm; according to the noise characteristic image, determining a training model; inputting a verification set noise adding image into the training model, and outputting a final training denoising model; and adding a test set noise adding image into the final training denoising model to test, and obtaining a denoised image, thus greatly improving the denoising efficiency and the denoising effect.

The invention provides a point cloud data de-noising method and system. The method includes the following steps that: three-dimensional point cloud data are acquired; the three-dimensional point cloud data are projected onto a two-dimensional plane, so that two-dimensional point cloud data can be obtained; and near-neighbor de-noising is performed on the two-dimensional point cloud data. According to the point cloud data de-noising method and system provided by the embodiments of the invention, the three-dimensional point cloud data are projected onto the two-dimensional plane, so that the two-dimensional point cloud data can be obtained, and namely, the three-dimensional point cloud data are converted into the two-dimensional point cloud data, and therefore, the data amount of the point cloud data can be decreased; and near-neighbor de-noising is performed on the two-dimensional point cloud data; and therefore, compared with a mode according to which near-neighbor de-noising is performed on three-dimensional point cloud data, the method can assist in decreasing the amount of data for calculation and improve de-noising efficiency.

The invention provides an address event stream data denoising method for a dynamic vision sensor, which mainly solves the defect of denoising address event stream data output by the dynamic vision sensor in the prior art, and comprises the following steps of: constructing a probability undirected graph model; Obtaining an energy function of the probability undirected graph model; Performing eventregion division on the address event flow data; And carrying out denoising processing on each event area. According to the invention, the correlation between events in address event flow data output by the dynamic vision sensor is utilized; wherein the construction comprises event line address information; a probabilistic undirected graph model of column address information and activation time information, The address event stream data is divided into a plurality of event areas for denoising, so that the integrity of the address event stream data information output by the dynamic vision sensoris ensured, the denoising efficiency of the address event stream data is improved, and the method can be used for denoising the address event stream data output by the dynamic vision sensor.

The invention relates to a single-photon laser height measurement data denoising method and device based on density statistics. The method comprises the following steps: 1) obtaining single-photon laser altimetry data; 2) performing data preprocessing; 3) carrying out photon density calculation by using an ellipse search area whose direction and size change adaptively with the gradient; and 4) adaptively calculating segmented threshold separation signal photons and noise photons by adopting an OTSU method, and removing discrete noise points by adopting a K nearest neighbor distance. Compared with the prior art, the method is not influenced by non-uniform background noise photons, and is high in adaptability, the direction of the search ellipse can be changed adaptively along with the slope, the size can be changed, signal photons and noise photons of each segment are separated by using the segmentation threshold, and a small number of discrete noise points left after rough denoising are further removed; and the method is complete in signal extraction, high in denoising precision and high in efficiency.

The invention discloses an image processing method and device, a mobile terminal video processing method and device, equipment and a medium. The image processing method comprises the steps of determining an image attribute of each image block in a to-be-processed image; determining a filtering model corresponding to each image block according to the image attributes; and filtering the image blocksthrough the corresponding filtering models to obtain the filtered image of the to-be-processed image. According to the embodiment of the invention, the image attribute of each image block in the to-be-processed image is quickly calculated, so that the filtering processing speed of the image blocks is increased, the real-time filtering of the image is further realized, and the definition of the video is optimized by performing adaptive filtering on each image block through the corresponding filtering model.

The invention discloses a method for removing super-high-density salt-and-pepper noises of an image, and mainly at solving the problem that a present method cannot be used to filter super-high-density salt-and-pepper noises. The method comprises the realization steps that (1) an extremum method is used to detect noise points; (2) values of the noise points are substituted by the mean value of non-noise points of a four-point template in a 3*3 window, recursion is carried out for three times, values of the noise points are substituted by the mean value of non-noise points of a crossing template in the 3*3 window, recursion is carried out for three times, whether all the noise points are processed is determined after each time of recursion, if no, the window is increased, and the four-point template and the crossing template are used for identical recursion mean value filtering till a 11*11 window is reached; and (3) if the noise density is greater than 0.85, a 3*3 crossing window or a full-domain window is used to continue mean value filtering, and the super-high-density noises are removed. The method of the invention has the advantages of low computing complexity, large noise removing range, and capable of inhibiting the super-high-density salt-and-pepper noises effectively.

The invention provides a signal processing method. The method includes: acquiring a to-be-denoised signal; detecting first waveform positions in the signal; dividing the signal into a plurality of signal segments according to the first waveform positions, wherein each signal segment includes at least one first waveform position; respectively judging and confirming signal quality of a signal in each signal segment; respectively determining a denoising method of denoising processing on the signal in each signal segment according to the signal quality of the signal in each signal segment; and using the denoising method of denoising processing on the signal in each signal segment for respective denoising processing on the signal in each signal segment. By adopting the above technical solution,matching of the denoising methods and changes of the signal quality can be ensured, denoising processing on the to-be-denoised signal is enabled to be more targeted, and denoising efficiency is improved.

The invention discloses an adaptive smooth filtering method. The method comprises the steps of acquiring a pixel point of an image to be denoised; and judging whether the pixel point is an obvious noise point or not; if the pixel point is an obvious noise point, acquiring a similar point set of the obvious noise point, and then estimating a pixel value of the obvious noise point; and adjusting a pixel value of the pixel point if the pixel point is not an obvious noise point. The adaptive smooth filtering method disclosed by the invention starts from an angle that self-similarity exists in the image texture, a filtering operation is carried out on an irregular non-singular point, and the image can be enabled to carry out region smoothing under the condition that the loss in edge and minutiae is reduced as much as possible.

The invention discloses an image noise filtering method via median and mean value iterative filtering of a minimal cross window, and aims at solving the problems that a present image de-noising method is hard to protect edge details and set parameters during de-noising. The image noise filtering method is realized by the steps that (1) noise points are found via an extremum method, and a binary map whose size is the same with that of an image is constructed; (2) the noise density p is calculated; (3) if p < / = 0.525, the iterative median of the minimal cross window is used for filtering, namely, the median of non-noise points in the minimal cross window replaces the value of each noise point in the noise image, iteration is carried out for another two times, whether all the noise points are processed is determined after iteration in each time, and if yes, a de-noising result is output; and (4) if p>0.525, the iterative mean value of the minimal cross window is used for filtering in a way similar to that via the iterative median. The image noise filtering method has the advantages that it is not required to consider the size of windows, operation is easy, the de-noising efficiency is high, details of the image are kept effectively, and the method is more suitable for real-time application.

The invention discloses electronic equipment and an antenna apparatus. The electronic equipment comprises a circuit board, at least an electronic component arranged on the circuit board, a first antenna arranged at a first position of the circuit board and used for transmitting and receiving communication data signals between the electronic equipment and other electronic equipment, and a second antenna arranged at a second position different from the first position of the circuit board, and the second antenna can be used for coupling noise signals generated by the at least one electronic component.

The invention relates to a high-noise image denoising method based on a deep convolutional network. The method comprises the following steps: firstly, carrying out feature extraction on an image accommodating noise by adopting incremental expansion convolution, batch standardization operation and a Leakly ReLU function; then the image is recovered, and a mode of combining decreasing expansion convolution and a ReLU activation function is adopted; realizing separation of image noise and content by the network model through combination of residual learning and batch standardization operation; and finally, the optimal weight parameter of the network model is learned by solving the value of the minimized loss function (adopting different loss functions for different noise distributions). And finally, denoising the noise image by using the trained network model. The method can effectively remove the image noise in a high-noise environment, improves the visual effect of the image, and is better in practicality.

The invention discloses a Rayleigh wave seismic data noise removal method. The method comprises the following steps: constructing a deep convolutional generative adversarial network; preprocessing the Rayleigh wave seismic data to obtain real noise-free data; adding Gaussian white noise based on the noise-free data to obtain noise-containing data; adding real noise-free data and noise-containing data as training samples into the deep convolutional generative adversarial network for training, and performing iterative calculation to obtain a deep convolutional generative adversarial network model; and carrying out de-noising processing on other Rayleigh wave seismic data to be de-noised. According to the method, random noise removal is carried out on the multi-channel surface wave data based on the deep convolutional generative adversarial network in the deep learning field, high-quality denoised Rayleigh wave seismic data is obtained, manual parameter adjustment is not needed, the labor cost is reduced, the Rayleigh wave seismic data denoising efficiency is improved, and the denoising effect is better.

The invention discloses an SAR image rapid denoising method based on RSVD and histogram storage. The method comprises the following steps: firstly, carrying out logarithmic transformation on an SAR image; multiplicative noise is converted into additive noise; The SAR image denoising method comprises the following steps: firstly, carrying out non-local similar image block matching on an SAR image,then carrying out low-rank matrix approximation on a low-rank matrix composed of non-local similar image blocks by adopting random singular value decomposition, carrying out texture enhancement on theimage by adopting a gradient histogram storage method, and finally resetting the image blocks to realize rapid denoising of the SAR image. Experimental results on an MSTAR database show that comparedwith an existing method, the method provided by the invention has the advantages that the edge preserving index is obviously improved, and meanwhile, the denoising speed is increased by three times.

The invention discloses an image denoising method for adaptive equidistant template iteration mean filtering, aimed at addressing the problems of current adaptive methods, such as poor denoising effect and poor recovery quality. The method includes the following steps: (1) using an extremum method to determine noise spots; (2) conducting sliding rotation with the middle points on 4 edges of a filtering window as start points and 4 angular points as terminal points, taking 8 symmetrical points of the 4 edges of the filtering window to construct an equidistant template (the 4 middle points and 4 angular points on the filtering window are 8 value template special cases), and forming a first equivalence template, a second equivalence template......, and conducting recursion clipping mean filtering with the equivalence templates; (3) checking whether a noise point is completed after the completion of each filtering from 3X3, and if the noise point is not completed, enlarging the filtering window and stops until 7X7, and forming adaptive filtering; (4) if the noise point is not finished, adopting iteration filtering. According to the invention, the method effectively processes noise and at the same time can better protect image details, has a higher rate of using information and has a rapid denoising speed.

The invention relates to a system and a method for acquiring product parameters, wherein the system comprises a target comment word frequency statistical unit, a product comment word frequency statistical unit, a correlation calculating unit, an expression pattern extraction unit and a matching unit, wherein the target comment word frequency statistical unit is used for carrying out statistics to all the word frequencies of each noun in an object comment and obtaining a word table; the product comment word frequency statistical unit is used for carrying out statistics to the word frequency of each word in a product comment and simultaneously, viewing the number of clauses of adjectives occurred in the word; the correlation calculating unit is used for calculating the correlation of each word in the word table; the expression pattern extraction unit is used for extracting the expression pattern of each product parameter in the target product comment and recording expression pattern tables; and the matching unit is used for finding product parameter words with low frequency of occurrences by matching the expression patterns in the target comment sentences.

The invention discloses a Hankel matrix structure optimization method and a device, computing equipment and a storage medium, and the method comprises the steps: setting an initial value of an iteration parameter, and setting the effective order of matrix reconstruction as k = 1; constructing a Hankel matrix by taking different values of the matrix parameter L, reconstructing a signal through SVDand calculating a signal-to-noise ratio to obtain a relation curve of the signal-to-noise ratio and the matrix parameter L; detecting whether the relation curve is a characteristic signal-to-noise ratio curve; taking a first peak point coordinate of the characteristic signal-to-noise ratio curve as a value of a matrix parameter L, setting a matrix column number as L + 1, and setting a matrix row number as N-L; wherein N is the signal length, and N is greater than L. The method can effectively improve the signal denoising efficiency.

The invention discloses an iterative denoising method based on speech recognition and a chip. The method includes the following steps: S1, determining a target speech signal and a target confidence value thereof; S2, selecting noise data matching the target confidence value from a noise database, and making the noise data and the unmarked voiced frames in the target speech signal participate in apre-denoising process; S3, judging whether the pre-denoising result is greater than a predetermined threshold, going to S4 if the pre-denoising result is greater than the predetermined threshold, otherwise, going to S5; S4, marking the voiced frames corresponding to the pre-denoising result as denoised voiced frames in the target speech signal; and S5, judging whether the absolute value of the difference between the confidence value of the pre-denoising result and the target confidence value is smaller than a confidence threshold, marking the voiced frames corresponding to the pre-denoising result as denoised voiced frames in the target speech signal if the absolute value of the difference is smaller than the confidence threshold, otherwise, adjusting the target confidence value and returning to S2.

The invention relates to the technical field of three-dimensional measurement, and provides a point cloud feature-preserving denoising method and device, electronic equipment and a storage medium, andthe method comprises the steps: constructing a preset tree model according to point cloud data, and searching a neighborhood point set corresponding to a main data point in the point cloud data basedon the preset tree model; calculating a point normal vector of the main body data point and a centroid point of a neighborhood point set according to the neighborhood point set corresponding to the main body data point through a preset algorithm; calculating a vector and a projection distance of the vector on the point normal vector according to the centroid point and the neighborhood point set;and if the projection distance satisfies a threshold distance, recording a to-be-denoised point corresponding to the projection distance as a noise point, and updating the neighborhood point set basedon the noise point until the neighborhood point set of the updated main data point is completely denoised. According to the invention, small ball-shaped noise and strip-shaped small-segment trace convex noise around the point cloud main body can be denoised, and the denoising efficiency is improved.

The invention discloses a linear fitting method and system and a storage medium, and the method provided by the invention can delete noise points in batch, and greatly improves the denoising efficiency. When the data noise is removed, compared with a traditional method, the noise existing in the data can be rapidly deleted by utilizing the correlation of the data, and the fitting method still hasgood robustness on the premise that a large amount of noise exists in the data.

The invention discloses a speech recognition-based iterative denoising device and a cleaning robot. The iterative denoising device comprises a confidence coefficient value acquiring unit, a pre-de-noising unit, and a processing unit; the confidence coefficient value acquiring unit is used for determining target speech signals from speech signals acquired by a microphone array, and obtaining a target confidence coefficient value accordingly; the pre-de-noising unit is used for receiving the target confidence coefficient value obtained by the confidence coefficient value acquiring unit, and selecting noise data matching the target confidence coefficient value from a noise database, and then controlling the noise data and unmarked sound frames in the target speech signals to be subjected to pre-de-noising processing so as to obtain a pre-de-noising processing result corresponding to the noise data; and the processing unit is used for adjusting the target confidence coefficient value according to a relationship between the pre-de-noising processing result outputted by the pre-no-noising unit and a predetermined threshold value, and transmitting the adjusted result to the pre-de-noisingunit so as to achieve denoising processing on the unmarked sound frames.

The invention discloses a fuzzy clustering analysis method for detecting SAR image changes based on non-local means. The method is implemented through the processes of inputting a difference chart composed of two SAR images in a same region at different times; correcting pixels of the difference chart according to similarity measure indexes in a fast global fuzzy C-Means clustering (FGFCM) algorithm to obtain a local spatial information pixel matrix; performing non-local mean processing on the difference chart to generate a pixel matrix of non-local filtering waves; weighting and summing up the two matrixes and generating a complete pixel matrix; clustering the complete pixel matrix through the FGFCM algorithm to generate a change detection binary result image and complete the change detection of the two SAR images integrally. According to the fuzzy clustering analysis method for detecting SAR image changes based on non-local means, local spatial information and non-local mean information of images are considered simultaneously and combined organically, so that noise influences are overcome effectively and image details are kept in an image analysis clustering process, and accurate difference chart analysis results are obtained.

The invention provides an image denoising method and device, electronic equipment and a storage medium, and relates to the technical field of image processing, and the method comprises the steps: determining a first sampling point and a first weight in a first direction; according to the current pixel value of the to-be-processed pixel point, the pixel value of the first sampling point and the first weight, performing first filtering processing in a first direction to obtain a first pixel value of the to-be-processed pixel point; and determining a second sampling point and a second weight in a second direction, and performing second filtering processing in the second direction according to the first pixel value of the to-be-processed pixel point, the pixel value of the second sampling point and the second weight to obtain a second pixel value of the to-be-processed pixel point as a current pixel value of the to-be-processed pixel point. According to the invention, the image denoising efficiency is improved, the operation complexity is reduced, the image denoising capability is improved, and the image quality of ISP processing can be improved.

The invention provides a cable partial discharge high-precision positioning and noise removing method and device. The device comprises a shell and a display screen, a mounting groove is formed in thetop of the shell; the display screen is mounted in the mounting groove through a rotating shaft; a first circular groove is formed in the front side of the shell. A second circular groove is formed inthe center of the inner side wall, away from the open end, of the first circular groove. A limiting sliding rod is fixedly installed between the inner side walls of the two sides of the second circular groove. The cable partial discharge high-precision positioning and noise removing method and device are reasonable in design, a stable and efficient synchronization mechanism suitable for the underground environment is provided, a double-end communication mechanism which takes the cable as a propagation channel and does not depend on the external environment is established at the two ends of the cable, then time synchronization and positioning are achieved, and therefore the positioning precision is improved.

The invention discloses a mud pulse signal processing method and device. The method comprises the steps: carrying out the forced denoising of a mud pulse sampling signal through frame synchronizationwavelet denoising, and obtaining a denoised frame synchronization signal; carrying out peak detection on the denoised frame synchronization signal to obtain a frame synchronization peak detection result; determining an original instruction signal in the mud pulse sampling signal according to a frame synchronization peak detection result; forced denoising is conducted on the original instruction signal through instruction wavelet denoising, a denoised instruction signal is obtained, and the length of the mud pulse sampling signal is twice that of the instruction signal. According to the invention, frame synchronization wavelet denoising is used to carry out first-stage forced denoising on a mud pulse sampling signal; and then secondary forced denoising is carried out on the original instruction signal through instruction wavelet denoising to obtain a denoised instruction signal, the denoising effect of the mud pulse signal is improved, calculation can be simplified through frame synchronization wavelet denoising and instruction wavelet denoising, and the denoising efficiency of the mud pulse signal is improved.

The invention provides an image denoising method and system. The method comprises the following steps: acquiring a noise image; inputting the noise image into a denoising model to obtain a denoised image; wherein the denoising model performs feature extraction on the noise image through the multi-scale feature extraction block and the Ghost feature extraction block in sequence to obtain noise features, and the noise features are subtracted from the noise image to obtain a denoised image. The image denoising effect is improved, the image denoising efficiency is improved, training parameters are reduced, and the network structure training time is shortened.

The invention belongs to the technical field of computer vision, and particularly relates to a picture document blind denoising system, method and device. The system comprises: a preprocessing unit which is used for performing Gaussian blur processing, graying processing and binarization processing on a to-be-processed picture to generate a first processing result; a straight line detection unit which is used for carrying out straight line detection on the first processing result to generate a first detection result; a table detection unit which is used for carrying out table detection on thebasis of the first detection result and generating a second detection result; a text direction processing unit which is used for carrying out text direction detection on the basis of the second detection result to generate a third detection result, and processing the first processing result according to the third detection result to generate a second processing result; a stretch removing unit which is used for carrying out stretch removing processing on the second processing result to generate a third processing result; and a distortion removing unit which is used for carrying out stretch removing or distortion removing processing on the third processing result to generate a final processing result. And the method can be perfectly adapted to a deep network model for OCR.

The invention discloses an image denoising method based on external non-local self-similarity and improved sparse representation. The method comprises the following steps: (1) dividing an external clean image data set into block groups; (2) dividing the block group of the external clean image into an external smooth block group and an external texture block group; (3) learning external smooth block group priori; (4) learning external texture block group priori through a Gaussian mixture model; (5) dividing the noise image into block groups; (6) using external non-local self-similarity prior to guide the clustering of the noise image block group, and calculating the smooth block group ratio of the noise image; and (7) optimizing regularization parameters of the sparse representation model by using a smooth block group ratio, and respectively recovering the image blocks in each subspace according to the improved sparse representation model. By optimizing regularization parameters of the sparse representation model, the self-adaptability of the sparse representation model is improved, and the technical problems that in an existing image denoising method, the peak signal-to-noise ratio of a denoised image is low, and detail information is lost are solved.

The invention discloses an image denoising method, a storage medium and terminal equipment, and the method comprises the steps: inputting a to-be-denoised image into a trained image denoising model, and carrying out the denoising of the to-be-denoised image through the trained image denoising model, so as to obtain a denoised image. The image denoising model is obtained by performing deep learningon the denoising process of a training image set with multiple training image groups, the training image groups comprise a first image and a second image with the same image content, and the signal-to-noise ratio of the second image is greater than that of the first image. Therefore, the trained image denoising model is used for denoising based on deep learning of the training image set, so thatthe operation performance of the image denoising model can be improved, the time consumed by image denoising is reduced, and the image denoising efficiency is improved.

The invention discloses a twin network image denoising method and system based on Transform, a medium and equipment, and designs two twin networks to extract complementary features, so that the robustness of an obtained denoising device is stronger. Transform is applied to a twin network, saliency features are extracted, a foreground and a background are separated, noise is removed, and a clean image is predicted; a cross interaction mechanism is designed to improve the memory ability of the deep network, and the denoising performance is improved; according to the method, batch normalization, layer normalization, instance normalization, a Swsh function and a linear rectification function activation function component are used in the twin network, so that the learning ability of the denoising network is improved, diversified features can be extracted, the denoising effect is enhanced, and the denoising efficiency is improved. In addition, denoising is carried out only through a 12-layer network, the calculation cost of the network is greatly reduced, and the requirements of mobile equipment are met very well. And saliency features can be adaptively extracted according to different scenes, and the method has a blind denoising function and a relatively high practical application value.