30 results about "Gradient noise" patented technology

A receiver, system, and method for performing equalization. The receiver includes a multi-channel chip equalizer for receiving a plurality of receive baseband signals and restoring chip pulse shapes of a plurality of transmit baseband signals transmitted by a plurality of transmit antenna to produce a plurality of equalized output streams and a correlator for correlating the plurality of equalized output streams with a correlation signal to reduce gradient noise in the plurality of equalized output streams. The method of equalizing includes receiving a plurality of receive baseband signals and restoring chip pulse shapes of a plurality of transmit baseband signals transmitted by a plurality of transmit antenna to produce a plurality of equalized output streams and correlating the plurality of equalized output streams with a correlation signal to reduce gradient noise in the plurality of equalized output streams.

The invention discloses image reference-free quality evaluation method and system based on a gradient profile. The image reference-free quality evaluation system comprises a gradient profile extraction device, a blurring effect evaluation device, a ringing effect evaluation device and an integrated evaluation device, wherein the gradient profile extraction device is used for detecting input image edge points and extracting the gradient profile according to the edge points; the blurring effect evaluation device is used for measuring the blurring effect of images according to the gradient profile; the ringing effect evaluation device is used for measuring the ringing effect of the images according to the gradient profile; and the integrated evaluation device is used for fusing the blurring effect measurement and the ringing effect measurement to acquire a quality evaluation reference value of input images. Based on the invention, the quality evaluation can be carried out on various types of natural images; and especially, the invention has favorable robustness for partially-blurred images due to the defocusing, the motion and other reasons.

The invention relates to compressed domain video lens mutation and gradient union automatic segmentation method and system. In the invention, aiming at three factors of feature extraction, dissimilarity measure and segmentation rules affecting segmentation performance of lens, a video feature extraction method based on principal component analysis and a method for building a texture feature map with a compressed domain are provided; secondly, a dissimilarity measure method in a time domain with multi-scale is provided; a method based on adaptive threshold is provided according to parameters oflocal features which can effectively characterize lens change and adaptively determine the length spaced by an N-frame time domain through a dissimilarity measure operator spaced by a 1-frame time domain; and the dissimilarity measure operator and effective distinguishing rules are designed for mutation and gradient lens. The invention can effectively strengthen the robustness of interference ofa lens segmentation algorithm to a camera or object movement in the lens by fully considering the three factors affecting the lens segmentation performance, have strong noise resistant performance andlow false drop rate, be fast and accurate, and greatly enhance the segmentation performance of the lens.

The present invention includes a method and apparatus to correct for gradient field distortions. The invention is particularly applicable in moving table imaging where a single extended image is desirable. The invention includes acquiring MR data in motion in the presence of gradient non-linearities, transforming the MR data acquired into the image domain, and then applying a warping correction function to the transformed MR data. The warp-corrected MR data is then corrected for motion induced during the MR acquisition. The data may be acquired point-by-point, line-by-line, or another sub-portion of the entire MR data acquired, and processed to minimize the amount of motion correction needed. Based on table velocity or acquisition sequence applied, the data is partitioned based on a common motion correction factor, and after correcting for motion, the data is accumulated to build up a final image.

A Noise Cancellation Process for enclosed cabins is disclosed. According to one embodiment, an input audio source corresponding to sound received from multiple microphones situated equidistantly in both directions in a two dimensional plane, is converted to a digital signal via an analog to digital (A/D) convertor. The A/D converted audio is analyzed for content to identify ambient noise. The frequency, amplitude and phase of the identified ambient noise is subsequently determined. A Noise correction sound wave is generated with negative phase of that corresponding to the identified ambient noise. The noise correction sound wave is added to the identified noise to create a noise corrected sound.

The invention provides a method and apparatus for measuring and calculating train noise in a tunnel and a storage medium and relates to the technical field of high speed train noise prediction. The method for measuring and calculating train noise in the tunnel comprises the following steps: train noise sources in the tunnel are divided into wheel-rail sound sources in a lower part of the vehicle body, aerodynamic line sound sources in a middle part of the vehicle body, and a current collection system sound source in an upper part of the vehicle body; a longitudinal section where a middle pointbetween adjacent first and second carriages is positioned is determined as a virtual wall, and a virtual room is formed via enclosure of the virtual wall and a surrounding wall of the tunnel; a soundpressure level of superimposed noise of the wheel-rail sound sources, the aerodynamic line sound sources, and the current collection system sound sources at a prediction point is calculated based onthe virtual wall. According to the method for measuring and calculating train noise in the tunnel, a plurality of sound source on a train are regarded as point sound sources via arrangement of the virtual room, the sound pressure level is obtained after noise of the point sound sources is superimposed at the prediction point, and prediction accuracy is improved when a noise measurement and calculation quantity is small.

The invention discloses a reference-free image objective quality evaluation method based on gradient self-similarity. According to the process of the method, first, gradient filtering is performed on a to-be-evaluated distorted image to obtain an amplitude image of gradient information; second, information images in four directions are obtained according to the amplitude image of the gradient information, and a self-similarity image between the information image in each direction and the amplitude image is calculated; third, inter-pixel feature extraction method operation and intra-pixel feature extraction method operation are performed on the self-similarity images to obtain an inter-pixel feature graph and an intra-pixel feature graph; fourth, a histogram statistical method is adopted to perform statistical analysis on the inter-pixel feature graph and the intra-pixel feature graph; and last, support vectors are adopted to perform regression prediction on an objective quality evaluation predicted value of the to-be-evaluated distorted image according to histogram statistical feature vectors. The method has the advantages that influences of changes of image gradient self-similarity on visual quality can be fully considered, and therefore the correlation between an objective evaluation result and subjective perception can be effectively improved.

An intelligent noise reduction system (100) is provided. The system can include a gradient microphone (110) to produce a gradient speech signal, a correction unit (116) to de-emphasize a high frequency gain imparted by the gradient microphone, a Voice Activity Detector 120 (VAD) to determine portions of speech activity (701) and portions of noise activity (702) in the gradient speech signal, an Automatic Gain Control 130 (AGC) unit to adapt a speech gain (740) of the gradient speech signal to minimize variations in speech signal levels, and a controller (140) to control the speech gain appliedby the AGC to the portions of noise activity to preserve a speech to noise level ratio between speech activity and noise activity in the gradient speech signal.

A system and method for simultaneously reconstructing magnetic resonance images and correcting those imaged for gradient nonlinearity effects are provided. As opposed to conventional methods for gradient nonlinearity correction where distortion is corrected after image reconstruction is performed, the model-based method described here prospectively accounts for the effects of gradient nonlinearity during reconstruction and implements a spatial support constraint to reduce noise amplification effects.

The invention discloses a brain signal online denoising method in a nuclear magnetic resonance environment. The method comprises the steps of 1) conducting high-pass filtering on collected brain electrical signals; 2) conducting up-sampling on the filtered signals and synchronizing the brain electrical signals with markers sent by a fMRI device; 3) constructing a sliding window to construct a noise template Atau for preliminary denoising; 4) slicing a signal Sh mixed with noise into a slice with a length of T*w points, fitting the noise template with slice date through least squares to obtainfitting parameters ytau, and subtracting ytau*Atau from the Sh to obtain a signal Sr with residual artifacts; 5) conducting PCA on the signal Sr, sorting each component according to degree of relevance, taking the largest m components as an optimal base betaj(j=1,2,...,m) of gradient noise, fitting the optimal base with the Sr through the least squares to obtain fitting parameter aj(j=1,2,...,m),subtracting N from the Sr, and conducting denoising through the construction of a gradient noise template and denoising through the PCA to eventually eliminate the gradient noise. The brain signal online denoising method in the nuclear magnetic resonance environment has the advantages that real-time denoising can be conducted on the signals to meet demands of data processing speed of an online experiment.

The invention discloses a document image binarization method based on stroke structural symmetry. The method includes the following steps: determining the gradient image of a document image, wherein the document image is a grayscale image; utilizing the OTSU method to conduct global binarization processing on the gradient image; based on the width of character strokes and the regional gradient symmetry, removing non-stroke gradient noise of the image which undergoes the global binarization processing, and determining a gradient image being regional gradient symmetrical; based on the gradient image being regional gradient symmetrical, determining a structural symmetry element image; based on the local intensity of foreground elements of the structural symmetry element image, filtering noise, and conducting local binarization processing based on voting strategy in combination with the document image. According to the invention, the embodiments increase the adaptability of character extraction from document image.

The invention discloses a Bayesian optimization exposure control method based on entropy weight image gradient, and relates to the technical field of vehicle-mounted cameras. According to the Bayesian optimization exposure control method based on the entropy weight image gradient, the entropy weight gradient is added on the basis of the gradient information of the original image, so that on one hand, gradient noise in the original image can be minimized, and on the other hand, a saturated mask can be established; therefore, the information loss of the original image is reduced to the maximum extent by reducing the overexposure or underexposure area, and the detail features of the original image are reserved; meanwhile, in order to avoid overlong exposure stabilization time of the camera, a Bayesian optimization algorithm is added in the exposure control method, and the optimal shutter time is obtained. The Bayesian optimization exposure control method is suitable for many computer vision target detection fields, such as scenes of lane detection, vehicle detection, pedestrian detection and the like.

The invention discloses an adversarial sample generation method and device. The adversarial sample generation method comprises the steps of dividing a region in an original sample into a target region and a non-target region; the target area is an area with high attention of a visual system, and the non-target area is an area with low attention of the visual system; determining gradient noise of the original sample, and dynamically distributing different step lengths for the target area and the non-target area respectively; respectively adding interference noise to the target area and the non-target area based on the step length and the gradient noise to obtain an adversarial sample corresponding to the original sample; the interference intensity of the non-target area is higher than that of the target area. According to the method, the disturbance noise amplitude and the noise information energy intensity can be increased under the condition that human eyes are not easy to perceive through small disturbance in the target area and high disturbance in the non-target area, so that the aggressiveness of the adversarial sample is improved, and the trained network model has higher robustness.

The invention discloses a federation recommendation gradient obtaining method and device, an intelligent terminal and a storage medium, and the federation recommendation gradient obtaining method comprises the steps: obtaining common objects and denoising objects; based on the model parameters, obtaining a first noise-containing gradient through each common object; based on the model parameters, obtaining a second noise-containing gradient through each denoising object; and based on the first noise-containing gradient and the second noise-containing gradient, eliminating the common object gradient noise and the denoised gradient noise, and obtaining a target gradient. In this way, the first noise-containing gradient and the second noise-containing gradient both contain corresponding gradient noise, and the scoring behavior of the user can be protected; meanwhile, corresponding gradient noise can be eliminated after the first noise-containing gradient and the second noise-containing gradient are obtained, and the target gradient without the gradient noise is obtained. Therefore, according to the scheme, gradient noise can be eliminated while user scoring behaviors are protected, andthe accuracy of the model in the federation recommendation process is improved.