33 results about "Speckle noise" patented technology

This invention provides a method for selecting strong scattering points in space-based ISAL systems by incorporating multi-dimensional features, relating to the field of ISAL image processing. This invention employs a multi-stage fusion processing strategy to form a complete link from image preprocessing to the final output of strong scattering points: pixel replication and extension are used to process convolutional boundary regions, significantly reducing boundary amplitude distortion caused by traditional zero-padding or periodic extension, and improving the signal-to-noise ratio of edge regions in the filtered amplitude map; the dynamic threshold generation process works synergistically with the preceding Gaussian kernel convolution preprocessing to suppress speckle noise while preserving the true amplitude characteristics of weakly scattering targets; amplitude-weighted averaging is used to calculate the centroid coordinates within the connected domain, overcoming the energy diffusion problem caused by the ISAL system's point spread function; transient interference points are eliminated based on the peak-to-mean ratio of multi-frame amplitude sequences, and the spatial distribution characteristics of the centroid position's spatial amplitude value and density constraints are combined to ensure that the output set of preferred scattering points simultaneously satisfies temporal stability, spatial saliency, and uniform distribution.

The application relates to a laser speckle image denoising method and device, which combines the complementary advantages of a laser intensity image and a laser phase image, extracts and fuses a contour feature map of a laser speckle image and a curvature feature map and a normal vector feature map of a laser phase image, provides a neural network with abundant geometric context, and further adds geometric feature constraints in the neural network to form a hierarchical system. The synergistic effect of the geometric constraints enables the denoising model to accurately maintain the edge contour, local surface morphology, surface orientation and overall topological structure of the image when removing speckle noise, significantly improves the geometric structure integrity of the denoised image, and effectively solves the problem that denoising and structure preservation are difficult to balance and diagnostic information is lost in the background technology. The application provides a more reliable and accurate solution for medical imaging application scenarios, and has significant technical progress and application value.

The present specification relates to a technique by which a hologram generation device generates a hologram with reduced speckle noise by optimizing a carrier according to a specific distance. A method by which a hologram generation device generates a hologram, according to an embodiment of the present specification, comprises the steps of: setting a carrier by initializing an arbitrary complex field having a random phase and a random amplitude; removing a specific band component from the frequency band of the set carrier so as to transform the set carrier; propagating the transformed carrier for each specific distance, which is a depth plane in which a target object is present; setting a loss function for each specific distance on the basis of a target uniform amplitude at the specific distance and the amplitude of the carrier propagated to the specific distance; and updating the propagated carrier for each specific distance on the basis of the loss functions set for each specific distance.

A SAR image change detection method based on a dual-stream hierarchical fusion engine network includes the following steps: A dual-stream hierarchical fusion encoder is constructed to collaboratively extract global and local features from SAR images at different time phases. The enhanced features are then used in a hierarchically aware U-shaped decoder to achieve hierarchical modeling and refined representation of the difference features. A frequency-domain channel attention mechanism with fused spatial weights is designed to enhance channel selectivity in the frequency domain and strengthen the response to change regions in the spatial dimension. A noise-resistant weighted loss function based on the difference map ablation coefficient is constructed to adaptively adjust the loss weights of each region, achieving noise region suppression and change region enhancement, effectively mitigating gradient bias caused by class imbalance and speckle noise. Training includes a dual-stream hierarchical fusion encoder and a U-shaped decoder network; SAR images from different time phases are input into the trained dual-stream hierarchical fusion engine network, which outputs a change detection binary map.

The application relates to the technical field of medical ultrasonic imaging and digital image processing, in particular to an ultrasonic elasticity image denoising and enhancing method, which comprises the following steps: acquiring original ultrasonic elasticity imaging data to be processed, and constructing a structure tensor matrix reflecting local texture directions; solving the anisotropic coherence degree of each pixel point, obtaining a structure attribute discrimination result, and constructing a structure confidence atlas; collecting the point spread function characteristics of an ultrasonic imaging system, establishing a speckle noise statistical model; obtaining a local smoothing coefficient of the first iteration, and performing anisotropic diffusion correction on an initial elasticity distribution matrix to obtain a first modified elasticity image matrix, and updating the structure confidence atlas to obtain a final denoised and enhanced target elasticity image; the application effectively suppresses multiplicative granular speckles, eliminates the ladder effect or artifacts easily generated by conventional methods, and significantly improves the image signal-to-noise ratio.

This invention discloses a lightweight remote sensing small target detection method based on edge enhancement and multidimensional attention. First, multi-scale feature extraction based on stacked phantom-guided reparameterization is performed on the remote sensing image. Frequency-domain guided gated edge enhancement is employed, and global context modeling is performed on the enhanced feature map based on a multi-head self-attention mechanism. After residual connection and layer normalization, the enhanced feature map is fed into a feedforward network for nonlinear transformation, outputting globally encoded high-dimensional features. A target query based on self-attention information interaction is then performed, interacting the target query with the high-dimensional features based on cross-attention. Lightweight multidimensional attention is then used to perform channel and spatial weighting on the interacted high-dimensional features, mapping them to target category and bounding box coordinates via a prediction head, outputting the final detection result. This method addresses problems such as complex backgrounds, large target scale differences, low pixel ratio of small targets, and inherent speckle noise in synthetic aperture radar images.

The present application relates to the technical field of image processing, in particular to an optical SAR ship re-identification method based on frequency domain adaptive enhancement and topological graph alignment, which solves the technical problems of SAR image speckle noise interference and geometric distortion in cross-modal ship re-identification by constructing a frequency domain adaptive enhancement module (FAAM), a modal decoupling feature extractor (MDFE) and a topological graph alignment module (TGAM); the FAAM adaptively mixes the amplitude information of the SAR image and the phase information of the optical image in the frequency domain to generate an enhanced image; the MDFE decouples the image to extract pure shared identity features; the TGAM models the local components of the ship as graph nodes and captures stable spatial topological relationships between cross-modal through a graph convolution network to realize geometric alignment; the present application combines frequency domain analysis and graph topological modeling to significantly improve the precision and robustness of cross-modal ship re-identification under complex sea conditions.

This invention discloses a training method, generation method, and apparatus for a generative model of optical coherence tomography (OCT) images, relating to machine learning. The training method includes: acquiring and preprocessing multiple samples under different illumination levels; inputting a first OCT image into a generator network in a preset model to generate a reconstructed image; inputting the reconstructed image and a second OCT image into a discriminator network to obtain the discrimination results corresponding to the reconstructed image and the second OCT image; calculating the generator loss of the reconstructed image and the second OCT image, and calculating the discriminator loss based on the discrimination results; updating the preset model using the loss value to obtain a generative model for OCT images. Based on this, this invention enables low-configuration imaging systems to reconstruct images with no speckle noise, high contrast, and high signal-to-noise ratio while rapidly generating low-quality images, according to the model.

A synthetic aperture quantitative phase imaging method based on a hybrid of digital holography and Fourier ptychography. A programmable LED array which is annularly arranged is mounted on an off-axis digital holographic system for providing oblique illumination at varying angles; accurate noise-containing low-frequency information of an object is obtained by means of digital holography and is used as initial estimation; and then, high-frequency information of the object is reconstructed by means of a Fourier ptychographic algorithm based on intensity measurement, so as to realize synthetic aperture quantitative phase imaging. Using the accurate low frequency provided by holography as initial estimation ensures the accuracy of phase retrieval and wavefront aberration reconstruction, without strictly meeting matching illumination conditions, and reduces the number of intensity graphs required by Fourier ptychography; and a non-interferometric phase retrieval method based on Fourier ptychography improves the imaging resolution to an incoherent diffraction limit, and also effectively solves the problem of inherent speckle noise in digital holography.

The application discloses a SAR remote sensing image ship detection method based on a double-main-stem network, relates to the technical field of remote sensing image processing and target detection, and aims at the problem of low accuracy of existing synthetic aperture radar ship detection. The double-main-stem architecture design in parallel with CNN and the transformer is adopted, the CNN main stem is focused on local detail feature extraction of the ship to resist the blurring influence of coherent speckle noise on the edge, the transformer main stem captures global semantic information to accurately distinguish the ship from the background such as sea clutter and islands, the features are complementary to each other, and thus the accuracy of synthetic aperture radar ship detection is improved.

The application relates to a SAR image water body submergence range change detection method and device for a flood scene. The method comprises the following steps: acquiring a training sample set containing pre-disaster and post-disaster SAR sample images and water body mask data, performing coherent speckle noise suppression and contrast enhancement preprocessing on the sample images, inputting the preprocessed images into a backbone double-branch twin network, extracting and interacting multi-scale features to generate double-time multi-scale feature maps, performing upsampling, splicing and channel space attention enhancement on the multi-scale feature fusion unit to obtain an optimized fusion feature map, generating a prediction result by a detection head, generating an intermediate heat map by a deep supervision unit, training a model to convergence by combining mask data to calculate a loss, inputting a to-be-detected image into the model after preprocessing to obtain a water body submergence range change detection result. The method can accurately extract the water body submergence range of a complex flood scene SAR image, effectively suppress noise, strengthen feature fusion, and improve the detection precision of the model to adapt to the real-time demand of flood emergency monitoring.

The application relates to the technical field of image processing, in particular to a SAR image denoising method and system. The method first extracts a speckle representation of a SAR noise image by using a twin encoder, and projects the speckle representation to an embedding space by a double-branch restoration network to generate low-frequency information embedding and high-frequency information embedding. Then, a double-branch attention network is constructed, and a loss function of the network is optimized by using the loss of the low-frequency and high-frequency information embedding, and then the network is trained by using the optimized loss function. Finally, the trained double-branch attention network is used for denoising the SAR noise image. The method effectively solves the technical problem that the denoising effect is poor due to the loss of edge details in the SAR image denoising process, can effectively suppress the speckle noise while reliably retaining the edge, texture and other key details of the image, and significantly improves the denoising quality.

The application relates to the technical field of remote sensing image processing, and discloses a SAR image target detection method based on global-local attention and multi-scale fusion enhancement, which comprises the following steps: acquiring a SAR image, inputting the SAR image into a continuous denoising attention network to perform speckle noise suppression processing, obtaining a denoised image, inputting the denoised image into a backbone network of YOLOv10 to extract a multi-scale feature map, inputting the multi-scale feature map extracted by the backbone network into a neck network, using an SAR perception multi-scale difference fusion module in the neck network to perform multi-scale feature difference extraction and fusion enhancement, obtaining an enhanced feature, using a weighted IoU loss function as a bounding box regression loss to optimize a predicted bounding box, and outputting a target detection result. The application can comprehensively improve the robustness and accuracy of image target detection, reduce the interference of speckle noise on subsequent detection, significantly enhance the model detection capability, and effectively improve the positioning accuracy of a rotating target and the convergence speed of the model.

The application provides an unbalanced SAR image recognition method and system based on a ring-shaped large-margin Gaussian mixture loss, and comprises the following steps: based on the characteristic that the coherent speckle noise in a measured SAR image conforms to a Gamma distribution, a training set image and a test set image with noise are constructed; a convolutional neural network for an unbalanced data set is constructed, and a multi-task loss function composed of a large-margin Gaussian mixture, a ring loss, an Euclidean loss and a total variation is constructed; based on the multi-task loss function and the training set image, a convolutional neural network model is trained to obtain trained network model parameters; based on the trained network model parameters, a test image set is recognized to obtain a recognition result as a synthetic aperture radar image target recognition structure; compared with the prior art, the application has the advantages of high recognition accuracy and strong noise resistance, and can be widely used in the field of image processing technology.

The application provides a SAR image denoising method based on space-frequency domain information aggregation, uses a space-frequency domain information aggregation network architecture, the network architecture combines the features of the space domain and the frequency domain, extracts the space domain information and the frequency domain information of the image through a double-branch structure respectively, and fuses the two complementary features through a double-domain aggregation module, so that the speckle noise in the SAR image is effectively removed, and the image details are restored. Specifically, it includes a space domain information branch, a frequency domain information branch and a double-domain aggregation module. The space domain branch is used to capture local structure details, the frequency domain branch is used to extract global context information, and the double-domain aggregation module fuses the features from the space domain and the frequency domain to improve the speckle removal performance. The application is superior to the existing most advanced method under multiple noise levels and performs well in SAR image speckle removal. The application effectively combines local space features and global frequency features, thereby achieving better speckle removal effect.

The application discloses a SAR image speckle noise suppression method based on sub-aperture decomposition and belongs to the field of radar image processing. The method comprises the following steps: performing sub-aperture segmentation processing and modulus quantization processing on single-view complex image data of a SAR image respectively, and sequentially obtaining sub-aperture images and a full-aperture image of the single-view complex image data; performing logarithmic transformation on an original three-dimensional tensor composed of the sub-aperture images and the full-aperture image to generate a noisy three-dimensional tensor; establishing a denoising objective function model containing a global low-rank regularization term, a non-local denoising regularization term and an edge preserving regularization term according to the noisy three-dimensional tensor, and performing iterative updating on the denoising objective function model until a preset maximum iteration number is met, and finally outputting a final denoised image with suppressed speckle noise. The application can improve the imaging quality of a SAR image after denoising processing.

This invention relates to the field of image data processing technology, specifically to an OCT-based image choroid segmentation system and method. The scheme performs discrete wavelet transform on the input image to separate low-frequency and high-frequency components; extracts low-frequency global context features and high-frequency boundary direction features through a dual-branch network; in the frequency domain cross-attention module, features are aligned using high-frequency boundary direction features as query keys and low-frequency global context features as keys and values ​​to suppress speckle noise interference and output fused features; the decoder receives the fused features and outputs a choroid segmentation mask. This invention solves the problem of the contradiction between speckle noise suppression and boundary preservation in existing spatial domain processing, filtering out noise while retaining weak high-frequency boundary information, thus improving the accuracy of choroid segmentation boundary localization.

The invention discloses a false target identification method and device based on low-rank sparse decomposition shadow extraction, and belongs to the field of synthetic aperture radar false target identification. Aiming at pain points with inaccurate shadow extraction and poor recognition robustness caused by speckle noise, the method comprises the following steps of: firstly, dividing regions of interest for a deception jamming SAR (Synthetic Aperture Radar) image; then, the CA-CFAR is used for eliminating strong scattering points; converting the image into a low-rank-sparse matrix through translation reconstruction, so that the shadow is in a low-rank state and the noise is in a sparse state; a residual threshold is adopted to drive low-rank sparse decomposition adaptive separation, and sparseness priori is not needed; oTSU binaryzation and morphological processing are carried out to obtain a preliminary shadow; according to a target-shadow fixed geometric angle, clustering correction is carried out, false detection is eliminated, and a final shadow is obtained; and finally, completing identification according to the condition that no shadow is a false target. According to the method, shadow and noise decoupling is realized under the single-channel condition, and the extraction precision and the recognition robustness are remarkably improved.

The application relates to the technical field of concrete ultrasonic flaw detection image processing, and discloses a concrete flaw detection ultrasonic image speckle noise suppression method and system based on multi-modal signal processing, which is characterized in that the original ultrasonic signal is subjected to adaptive white noise injection and EMD decomposition, non-aliasing IMF components are obtained through modal alignment and averaging, and high-frequency noise dominant components and medium-low-frequency effective signal components are divided according to the kurtosis criterion; subsequently, the high-frequency components are subjected to VMD optimization, the frequency band is accurately divided by constructing a variational constraint model and iteratively solving; the improved adaptive wavelet threshold function and the hierarchical hybrid threshold strategy are adopted for the medium-low-frequency components, so that the noise is accurately filtered out and the effective signal is retained; finally, the de-noised components of various frequency bands are fused to generate a one-dimensional time domain signal, the residual noise points are removed through median filtering after being restored into a two-dimensional image, and the de-noised enhanced image is output. The application realizes accurate separation of the noise and the effective signal, and completely retains the fine defect features in the concrete.

This invention relates to the field of image correction technology, specifically to an ultrasound image correction system based on dynamic range. The system includes a cepstral conversion module, used to extract ultrasound image scan line signals from an ultrasound probe, extract depth level sequences along the axial depth direction and convert them into logarithmic spectral level parameters to obtain a cepstral transform coefficient matrix, extract the amplitude and inverse frequency values ​​corresponding to the extreme points of equally spaced isolated peaks within the cepstral transform coefficient matrix to form feature vector parameters, and generate reverberation period peak fractions. In this invention, an adaptive truncation limiting variable is calculated based on the local dynamic range parameter set to extract the original coordinates of the center pixel grayscale level data, truncate and assign values ​​to establish a corrected output image matrix, and implement dynamic response precise truncation for speckle intensity in different regions. This effectively suppresses high-frequency speckle noise while maintaining the sharpness of edge contours, improving image contrast and spatial fidelity of the underlying data.

This invention discloses a polarimetric SAR change detection method and system based on a Siamese attention complex convolutional neural network. It addresses the issues of poor quality difference maps generated by speckle noise in PolSAR images, and the problem that current real-number networks for SAR image change detection can disrupt the integrity of the complex structure of PolSAR data. This invention utilizes a Siamese network architecture to construct a more robust multi-scale feature difference map to suppress the influence of speckle noise. Furthermore, it combines a complex network to maintain the integrity of the PolSAR data structure and reduce polarimetric information loss. A corresponding complex attention module is also designed to enhance the identification of changed regions. Finally, the change detection results are obtained by decoding the multi-scale feature difference map. By using a Siamese attention complex convolutional neural network, the invention enhances the utilization of polarimetric information while suppressing the influence of speckle noise, thereby improving the accuracy of PolSAR change detection.

This invention discloses an automatic method and system for identifying supersonic flow field shock waves from high-noise schlieren images. The method includes: converting the input schlieren image sequence to grayscale, cropping, and filtering to reduce noise and remove invalid background regions; performing region growing based on image gradient calculation and direction consistency to construct line support regions and obtain candidate line segments through weighted rectangle fitting; eliminating false edges by combining angle thresholds and exclusion region rules to obtain a set of shock wave segments that conform to physical characteristics; identifying the dominant shock wave segments through angle statistics and comprehensive scoring, and automatically outputting their angle, position, and other parameters. This method can achieve high-precision, batch-based automatic identification of supersonic flow field shock waves in high-noise environments with uneven illumination and significant speckle noise. The system has real-time processing capabilities, supports multi-frame shock wave structure synthesis and visualization, and provides an efficient and reliable technical means for quantitative shock wave analysis and flow field evolution research in aerospace inlets and wind tunnel experiments.

This invention provides an adaptive high-frequency signal filtering and enhancement method for SAS imaging of subsea pipelines, relating to the field of signal processing. The method includes: performing acoustic propagation compensation on the original high-frequency echo signal; obtaining logarithmic domain distance data through range pulse compression and homomorphic logarithmic transformation; traversing the resolution cells to construct a joint discriminant factor; adaptively determining the global noise threshold set and classifying the resolution cells into noise-dominant regions, edge-textured regions, or flat regions; applying differentiated filtering strategies for different region types for range filtering; and combining the results with inverse exponential transform and azimuth-matched filtering to output an enhanced image. This invention achieves accurate region attribute identification through the joint discriminant factor, and by combining adaptive threshold segmentation and differentiated filtering, effectively suppresses speckle noise while preserving the pipeline edge structure, significantly improving the imaging quality of subsea pipelines.

The application provides a medical ultrasound database construction method, and relates to the technical field of ultrasound database construction, which comprises the following steps: acquiring patient basic information, assigning a unique patient number and examination serial number to each patient; deploying a work list server and transmitting image files to an image server; extracting the examination serial number, establishing an examination serial number table to record examination summary information; using an adaptive filter to suppress speckle noise of each frame, and obtaining a feature vector through a deep convolutional neural network; calculating the local content stability, forward probe motion amplitude and content novelty of the i-th frame, and obtaining a diagnosis behavior state sequence through a state determination rule; establishing a sequence mapping table; establishing an image table; and establishing a disease index table recording the mapping relationship between disease categories and examination serial numbers, so as to realize image sample retrieval based on disease categories. The application can realize standardized storage and efficient retrieval of ultrasound multi-frame sequence data, and provide structured data support for artificial intelligence model training.

The invention discloses a digital holographic microscopic imaging coherent noise suppression network model and method based on deep learning. The network model adopts a double-branch encoder-single decoder structure, and a double-branch encoder extracts local details and global noise features of an image in parallel; the innovative double-branch intensive attention fusion module fuses multi-scale features through an adaptive weighting and enhancement mechanism; and the decoder integrates information through jump connection and reconstructs a clear image. The training method generates a high-fidelity data set based on a physical simulation system. According to the method, speckle noise and parasitic fringes can be efficiently suppressed, the phase details of the object are effectively reserved while the peak signal-to-noise ratio and the structural similarity index are remarkably improved, and the method has excellent generalization ability and calculation efficiency and is suitable for real-time high-quality imaging of a digital holographic microscopy system.

The application relates to a laser speckle image enhancement method and device based on deep learning, which simultaneously utilizes an original image and a second image subjected to logarithmic bilateral filtering, and extracts high and low frequency components thereof, thereby providing multi-source information for subsequent feature aggregation, and being capable of more comprehensively capturing image information and overcoming the limitation of a single image source; then, a basic model pre-trained based on laser speckle training samples is introduced to extract speckle features from the first image, and the generation of high-frequency features and low-frequency features is effectively guided through domain knowledge representing laser speckles, so that high-frequency guided features and low-frequency guided features are obtained; the knowledge guiding mechanism enables the model to more effectively distinguish effective textures and noises in the enhancement process, avoids blind noise reduction, and maximally retains and restores key texture details while suppressing speckle noise, thereby providing a more reliable and accurate image basis for subsequent medical diagnosis and quantitative analysis.

To improve the data quality in detecting cardiac wall motion, regions of abnormal wall motion are detected from a view. The scan settings are then changed to focus scanning on each region, providing improved data such as data with more speckle being present. The scan settings may include changing an orientation of the scan plane, reducing out-of-plane motion, and / or increasing speckle content. The improved data is used to more accurately determine strain.

This invention belongs to the field of marine remote sensing technology, specifically relating to an automatic SAR ocean internal wave detection method based on a position-adaptive convolutional neural network. The method includes collecting and preprocessing SAR image data within a target area over a given time period; drawing internal wave mask images as ground truth labels based on the internal wave characteristics of the images; constructing a position-adaptive convolutional neural network; training the model using remote sensing images and ground truth labels to obtain an internal wave extraction model; and inputting SAR images from other time periods into the internal wave extraction model to extract internal wave information, thereby achieving accurate identification and mapping of the morphology and location of ocean internal waves. This invention is based on deep learning for ocean internal wave recognition, overcoming the difficulty of traditional image detection methods being susceptible to speckle noise and improving the fragmented recognition results caused by the inability of conventional convolutional methods to adapt to internal wave morphology.

The present application belongs to the technical field of image denoising, in order to solve the problems of weak generalization ability and restricted denoising of existing fundus image speckle noise suppression methods, an image denoising method and system based on physical information guidance are proposed, the registered average living fundus image and the false eye static average image are randomly fused into a feature map to train the speckle noise estimation network; in the training process, based on the estimation loss calculated by the mask area features of the false eye static average image with the prediction results of the speckle noise estimation network and the signal-to-noise ratio tending to be stable, and the noise distribution perception loss calculated by the feature maps corresponding to the false eye static average image with the prediction results of the speckle noise estimation network and the signal-to-noise ratio tending to be stable, the training of the speckle noise estimation network is guided. The present application uses the false eye static image with high signal-to-noise ratio as physical information to guide the speckle noise estimation network to realize high signal-to-noise ratio denoising under different noise distributions.

This application provides a target-oriented crop plot-level classification method and device for fragmented farmland. The method includes: using a U-TAE encoder-decoder backbone network to extract multi-scale features from the temporal data of multi-temporal remote sensing image sequences in the input information; using a parallel spatiotemporal self-attention module to enhance the temporal and spatial features of the features output from the encoding stage of the backbone network; using a parameterless plot aggregation operator to aggregate the fused features into a plot-level representation; performing plot-level classification on the plot-level representation; and back-broadcasting the classification results to the corresponding pixel location of the plot based on the plot ID map. This application captures long-range dependencies through U-TAE and spatiotemporal attention, achieves end-to-end plot-level classification through parameterless plot aggregation and bi-branch prediction, eliminates salt-and-pepper noise while maintaining accurate boundaries, suppresses SAR speckle noise, and achieves efficient and accurate identification of fragmented farmland.