14 results about "Noise Artifact" patented technology

The invention discloses a polarization degree and light intensity image fusion method based on bidirectional cross attention. The method belongs to the technical field of polarization imaging, computational imaging and image processing. The technical problems that in the prior art, dynamic balance between structural detail keeping and polarization saliency enhancing is generally difficult to achieve, and particularly under the condition of a complex background or a low signal-to-noise ratio, texture missing or noise artifacts are prone to occurring in a fusion result are solved. A fusion mechanism capable of establishing a two-way information interaction relationship between a light intensity image and a polarization degree image is provided, and further processing is carried out aiming at a noise problem, so that more stable and more reliable polarization fusion imaging is realized.

Blurring and noise artifacts in magnetic resonance (MR) images caused by off-resonant image components may be corrected with convolutional neural networks, particularly feed forward networks with skip connections. Demodulating complex blurred images with off-resonant artifacts at a selected number of frequencies forms a respective real component frame of the MR data and a respective imaginary component frame for each image. A convolutional neural network is used to de-blur the images. The network has a plurality of residual blocks with multiple convolution calculations paired with respective skip connections. The method outputs, from the convolutional neural network, a de-blurred real image frame and a de-blurred imaginary image frame of the MR data for each complex blurred image.

The invention discloses an X-ray and NeRF algorithm-based luneberg lens internal structure analysis method, and relates to the technical field of three-dimensional imaging of optical lenses, and the method comprises the following implementation steps: S1, configuring an RGB-X-ray combined imaging system, and carrying out data acquisition, and S2, carrying out the calibration of external parameters of a camera through a calibration plate: carrying out the imaging through a specially-made calibration plate, and carrying out the calibration of the external parameters of the camera through the calibration plate. The method comprises the following steps: acquiring a plurality of frames of RGB (Red, Green, Blue) images to obtain external parameters of an RGB camera under a calibration plate coordinate system, acquiring a plurality of frames of X-ray images to obtain an external parameter matrix of an X-ray camera about the calibration plate coordinate system, performing implicit representation on a continuous field in a lens by utilizing NeRF, mapping three-dimensional coordinates to differentiable density and radiation values, and calculating the calibration plate coordinate system. According to the method, high-fidelity reconstruction from multi-angle X-ray projection to a three-dimensional structure is achieved, detail information in the lens can be effectively reserved, meanwhile, the continuity and symmetry of the structure are guaranteed, and compared with a traditional voxelization or CT reconstruction method, the problems that noise artifacts, fractures and unstable numerical values are likely to be generated in the traditional voxelization method are solved.

The invention belongs to the field of computer vision, and particularly relates to a multi-mode CNN-Transform fused image tampering detection method, which designs a CNN and Transform double-flow parallel feature extraction structure, effectively combines the advantage of CNN at capturing fine local features and the advantage of Transform in capturing long-distance dependency relationship and global semantic information, and improves the accuracy of image tampering detection. Local details and global tampering features in the image can be sensitively perceived at the same time, so that the detection capability and stability of image tampering are effectively improved. Processing the noise domain image filtered by the SRM through a CNN (Convolutional Neural Network), and capturing local texture and noise artifact features; meanwhile, a Transform branch processes an RGB spatial domain image, extracts rich global semantic information, and realizes deep interaction and advantage complementation of two types of modal information through a feature fusion module. The BAFM module provided by the invention can deeply mine feature information of different scales and different spatial directions, and more accurate spatial attention features are generated, so that the sensitivity and expression ability of the network to tampered regions are improved.

This invention discloses a method for constructing an interpretable predictive model of pathogenic bacteria in chest CT scans, belonging to the field of artificial intelligence technology. The method includes: Step S1: acquiring a chest CT image dataset and pathogenic bacteria type labels, and constructing a pathogenic bacteria imaging knowledge graph using a medical knowledge base; Step S2: constructing an initial model, which includes: a noise-invariant encoder, a counterfactual causal intervention module, a knowledge-guided feature compensation module, an interpretability analysis module, and a classifier; Step S3: feeding the spatial attention weight map back to the noise-invariant encoder to update and iterate the anatomical pathological features. Based on the final features obtained after internal iterative optimization, the initial model is trained according to preset training rules until overall convergence. This invention effectively distinguishes between real lesions and noise artifacts in low-dose CT images through causal intervention and attention map difference-driven feature updates, improving the model's recognition accuracy and robustness.

The application relates to an image cognitive interpretation information density evaluation and promotion method, and belongs to the technical field of optical imaging application. The method is as follows: starting from the human eye cognitive mechanism and interpretation process, image interpretation quality influence elements and key parameters and the correlation between the two are extracted; the cognitive interpretation information density evaluation model is established by comprehensively considering the key parameters; based on the cognitive interpretation information density evaluation model, a processing algorithm is designed to realize fine promotion of image interpretation quality by taking into account the transmission function promotion, texture detail preservation and noise artifact suppression. The application solves the problem that the interpretation quality elements are not clear in the prior art, fills the technical gap of effective information content quantitative evaluation, breaks through the technical bottleneck that the traditional processing is easy to cause artifacts, forms a complete technical system from information density evaluation to interpretation quality promotion, improves the ground object classification precision and target recognition rate, and provides key support for intelligent application of optical remote sensing images.

The application discloses a kind of low-dose CT reconstruction methods of prior image and convolution sparse network in combination, belong to computer tomography technical field.The present application includes the following steps: first, using distinctive feature representation method obtains high-quality prior image;Second, design the convolution sparse network of feature fusion;Next, the error of reconstruction image and prior image is calculated, and iterative reconstruction gradient is estimated according to error;Then, the reconstructed image is fine-tuned using total variation constraint;Finally, the iterative reconstruction of low-dose CT is realized by module cascade form.The present application algorithm comprehensively utilizes the advantage of prior image of plan scanning and convolution sparse network, has explainable strength, reconstructed image quality is high, and has the advantages such as less noise artifact, has greater advantage in clinical tumor radiotherapy, can improve examination efficiency, reduce the radiation injury of patient non-target organ.

A system and a method for noise artifact mitigation in Time-of-Flight cameras is disclosed. A receiving module receives a depth frame of an image captured by the Time-of-Flight camera. A noise artifact detection module includes an artificial intelligence module detecting one or more regions affected by noise artifacts and output corresponding to a plurality of bounding box coordinates. A noise artifact verification module verifies whether the one or more regions detected are affected by the one or more noise artifacts or represents false positives using a noise artifact identification algorithm. A depth reconstruction module removes a false depth information from the one or more regions, estimate a corrected depth information for the one or more regions and employ a statistical based depth reconstruction algorithm with a region-specific reconstruction filter to replace the false depth information to generate a final denoised depth frame.

The present application relates to the technical field of image processing, more particularly, the present application relates to a kind of surgical image acquisition method and system based on image enhancement, comprising: obtaining the original image obtained by CT scanning, and the original image is preprocessed to obtain surgical image image;In the local neighborhood window of any pixel point in the surgical image image, the absolute value of the difference between the gradient amplitude of each pixel point and the average value of the gradient amplitude in the window is multiplied by the exponential function of the absolute value of the difference between the gray value of each pixel point and the gray value of the center pixel point in the window.The present application constructs the jump potential by analyzing the asymmetry of pixel neighborhood gradient, and calculates the structure confidence by tracking the potential change along the edge direction, so as to accurately distinguish the real anatomical edge from noise artifact.Furthermore, the structure confidence is used to adaptively adjust the Gaussian scale, while maintaining the contrast enhancement in the flat area, the Gaussian kernel is forced to shrink at the edge, which effectively eliminates the halo and improves the detail clarity.

The disclosure is a low-light image enhancement method based on double codebook reconstruction and wavelet refinement, which comprises: image color space conversion; double codebook collaborative color reconstruction: in the color branch, the bright codebook and the dark codebook collaborative reconstruction strategy is adopted, and the context perception module is combined to complete the color fidelity recovery of the extremely dark area; frequency decoupling wavelet illumination enhancement: in the illumination branch, the intensity map is decomposed into low-frequency component and high-frequency component through the frequency decoupling wavelet transform module, then inverse wavelet transform is carried out, the enhanced HV color map and the optimized intensity map are fused into HVI image, and the final RGB enhanced image is output through the inverse HVI converter. The embodiment adopts double codebook and mask fusion, guarantees the color stability of the normal illumination area, and improves the color discrimination degree of the extremely dark area; through the frequency decoupling wavelet transform and the cross-frequency guide, the illumination and the details are collaboratively optimized; the context perception module is used to suppress noise artifacts, enhance spatial consistency and robustness.

This application provides an ultrasound processing device, an ultrasound image processing method, and a medium, relating to the field of image processing technology. The ultrasound processing device includes a memory and a processor. The processor is used to: acquire an initial ultrasound image and convert the initial ultrasound image from the spatial domain to the frequency domain to obtain a spectral image; suppress the frequency values ​​of the spectral image according to the position and / or magnitude of each frequency value in the spectral image to obtain a suppressed spectral image; convert the suppressed spectral image from the frequency domain to the spatial domain to obtain a spatial image; and compensate for the target region in the spatial image based on the initial ultrasound image to obtain a target ultrasound image, effectively removing noise artifacts from normal tissues and improving the uniformity of brightness display in different regions of the image.

This invention discloses a non-invasive method and apparatus for acquiring electroencephalogram (EEG) signals, relating to the field of EEG signal acquisition technology. The method includes: reading the sensor distribution scheme of the acquisition sensor group and performing noise signal identification; performing partitioned clustering and establishing partitioned clustering results; acquiring partitioned EEG signal datasets; performing noise intensity discrimination; establishing noise artifacts for the corresponding partitioned EEG signal datasets; configuring hysteresis noise in the partitioned channels using the noise artifacts, performing partitioned suppression of the corresponding partitioned EEG signal datasets, performing cross-channel signal authentication, and outputting the EEG signal acquisition results. This invention solves the technical problems of insufficient accuracy in noise identification and poor suppression effects in the existing technology during EEG signal acquisition, resulting in low accuracy and reliability of EEG signal acquisition. It achieves accurate identification and suppression of noise in EEG signals, improving the accuracy and reliability of EEG signal acquisition.

The invention discloses a noise artifact dynamic identification method based on multi-scale feature modeling and deep learning. The method comprises the following steps: S1, continuously collecting complete noise signals in an environment; s2, preprocessing the digital signal; s3, identifying whether the short-time frames contain artifact noise or not and removing the short-time frames containing the artifact noise; s4, maintaining and updating an artifact signal pattern library and a deep learning recognition model; s5, post-processing is carried out on the noise signal after artifact noise identification is completed; and S6, calculating indexes such as a continuous equivalent sound pressure level for the post-processed noise signal. The invention discloses a noise artifact dynamic identification method based on multi-scale feature modeling and deep learning, and aims to realize accurate identification and removal of artifact signals in an industrial noise acquisition process through combination of multi-scale time-frequency feature analysis and a deep learning algorithm.

Systems and techniques for a reverberation cancellation framework include receiving a far-field audio signal from a far-field microphone array and a near-field audio signal from a near-field microphone array, where the far-field microphone array is a greater distance from an audio source than the near-field microphone array. The far-field audio signal and the near-field audio signal are synchronized. The far-field audio signal and the near-field audio signal are encoded to remove noise artifacts from the far-field audio signal and the near-field audio signal. The far-field audio signal and the near-field audio signal are decoded to output an output audio signal with the noise artifacts removed.