85results about How to "Smooth noise" patented technology

The present invention discloses a facial shape tracking method based on space-time cascade shape regression. The method comprises: detecting the first frame of a video through combination of a face detector configured to detect and register, initializing the whole system, and obtaining five facial feature points; assessing the five facial feature points to obtain similar transformation parameters (rotation, offset and scale) and face poses (left sides of the faces, right sides of the faces and the faces); and employing multi-view cascade shape regression to predicate the face shape of the current frame, when the registering result confidence is larger than a setting threshold, allowing the time sequence regression to set about tracking the face shapes, and when the registering result confidence is smaller than a setting threshold, starting a re-initialization mechanism to perform stable tracking of the face shapes. The facial shape tracking method based on space-time cascade shape regression is faster in convergence speed and higher in precision through the multi-view (the left side of the face, the right side of the face and the face) cascade regression, and faster in the face shape tracking speed and more accurate in the face shape tracking through the time sequence regression and the re-initialization mechanism.

The invention relates to an EEG noise elimination method based on dual-density wavelet neighborhood related threshold processing. At present, noise elimination is carried out on an EEG mostly by adopting classic discrete wavelet transform to be combined with a traditional threshold method, and defects exist in an existing noise elimination method with the combination of the classic wavelet transform and the traditional threshold method. The EEG noise elimination method comprises the steps: firstly collecting an EEG from a cerebral cortex, then using dual-density wavelet forward transform for conducting decomposition on the EEG to obtain multi-layer signal high-frequency coefficients, utilizing a neighborhood related threshold processing algorithm for contraction according to the partial statistics dependency of wavelet coefficients, and finally reconstructing the contacted wavelet coefficients to obtain signals with noise eliminated. According to the characteristics of the EEG and the characteristics of interference noise, the signal to noise ratio is used as an objective function, a grid optimum seeking method is adopted to seek the optimum in three adjustable parameters of the neighborhood related threshold processing algorithm, then the noise is effectively smoothed, and the detail features of the EEG are reserved.

The invention discloses a method, device and system for controlling a variable-frequency air conditioner. The method includes the steps of obtaining running frequencies of the variable-frequency air conditioner and noise signals corresponding to the running frequencies; generating a frequency and noise distribution curve of the variable-frequency air conditioner on the basis of the running frequencies and the noise signals; determining one or more target frequencies of the variable-frequency air conditioner according to the distribution curve, wherein the values of the noise signals corresponding to the target frequencies are not higher than the preset noise threshold value; controlling the variable-frequency air conditioner to run at the target frequency corresponding to the running request of the variable-frequency air conditioner. By means of the method, device and system for controlling the variable-frequency air conditioner, the problem that in the prior art, noise changes are large in the running process of a variable-frequency air conditioner system is solved, and the effects of ceaselessly controlling noises of the variable-frequency air conditioner and being low in noise are achieved.

The present invention provides a friction material having reduced noise after being left in cold conditions when used as a disc brake pad, brake lining, or the like, and copper components are essentially not included in the friction material. The present invention is: a friction material composition including a fiber base material, a binding material, and a friction adjusting material, the copper content of the friction material composition being no more than 0.5% by mass, and the friction material composition containing fluorine-based polymer particles; and a friction material obtained by molding the friction material composition.

Apparatus, systems, and methods to recognize features on bottom surfaces of containers on a container production line, detect defects in the containers, and correlate the defects to specific production equipment of the container production line, based in part on the recognized features. The system includes imaging apparatus, programmable processing devices, and controllers. The methods include imaging techniques and estimation techniques.

The invention relates to a soya bean milk maker circuit with motor soft start function, which comprises a motor, a silicon-controlled switch which is connected with the motor at an output end, a trigger circuit unit connected with a control end of the silicon-controlled switch, a single chip microcomputer connected with the trigger circuit unit, a zero-crossing detecting unit connected with the single chip microcomputer and a power and voltage detecting unit connected with the single chip microcomputer. An input end of the silicon-controlled switch is connected with one pole of a power, with the other pole of the power being connected with the motor at one end that is not connected with the silicon-controlled switch; the power and voltage detecting unit is connected with two poles of the power; the single chip microcomputer is connected with the power through a power step-down circuit unit. The soya bean milk maker circuit with motor soft start function of the invention controls the motor under the condition that the trigger circuit unit controls the silicon-controlled switch, thereby realizing soft start, and when the motor is started, shock impact of the motor on electric network is reduced and sudden noise caused by starting the motor becomes stable, thus reducing acoustical irritation to human body; in addition, the motor is provided with various voltage for regulating rev of the motor, thereby meeting requirements for different crushing degrees determined by bean materials and users, reducing the noise of the motor, prolonging the service life of the motor and increasing discharging rate.

The invention discloses a non-local mean denoising method based on joint similarity, belonging to the technical field of image processing. The method mainly solves the problem of inaccurate calculation of weight in the existing natural image non-local mean denoising technology. The implementation process of the method comprises the following steps of: (1) setting a searching region for all the pixels in an input noise-containing natural image, averaging and pre-selecting variance for the pixels in the searching region to obtain a similarity set; (2) calculating distance from the current pixelto all the pixels in the similarity set, and calculating the weight through a weight formula designed in the method; (3) performing weight average on all the pixels and corresponding pixel blocks in the similarity set according to the calculated weights of all the pixels in the similarity set to obtain gray values of image pixels and pixel blocks after pixel modification. The method provided by the invention is superior to other denoising methods in overall performance, can keep details of edges, textures and the like of the natural image while smoothing the noise better, and can be used for denoising treatment for the natural image.

A processing system generates an output signal which includes desired signal components, and reduces or eliminates tonal noise. The output signal may be provided to any subsequent signal processing system, including voice recognition systems, pitch detectors, and other processing systems. The subsequent processing systems are less likely to mistake tonal input signal noise for desired signal content, to needlessly consume computational resources to analyze noise, and to take spurious actions induced by the tonal noise.

The invention provides an embossment modeling method based on a single photograph. The embossment modeling method comprises steps of: converting the single photograph into a grayscale map; determining the outline of the grayscale map; determining the heights of pixels on the surface layer of the embossment according to the gray value of the grayscale map, constructing the top surface of the embossment and the side surface of the embossment; adding a bottom surface according to a requirement for the thickness and the shape of the bottom surface according to a process and on the basis of the outline; reconstructing the triangular patch of each surface; and generating a model file in a STL format. The method acquires the corresponding grayscale map by preprocessing the single color photograph, constructs the embossment model according to the gray value of the grayscale map, generates the STL-format model file which can be directly used for 3D printing, solves a technical problem of low degree of dissemination in the conventional three-dimensional reconstruction technology based on three-dimensional scanning and multiple photographs with different angles, and gives a more lifelike three-dimensional model.

The invention discloses an image denoising method based on Treelet transformation and minimum mean-square error estimation, mainly solving the problem that the existing denoising method is poor in denoising effect of the natural images corroded by Gaussian white noise. The image denoising method comprises the following steps of: (1) inputting an image to be denoised; (2) searching similar image blocks; (3) obtaining a basis matrix by Treelet transformation; (4) projection; (5) calculating the coefficient after denoising; (6) calculating a similar image block matrix after denoising; (7) calculating the similarity; (8) valuating the matrix; (9) judging whether all the image blocks in the image are processed completely, if so, carrying out a step (10), and if not, carrying out a step (12); and (10) outputting a result. The image denoising method has the advantage that the denoising effect of the natural images containing Gaussian white noise, can restore the original characteristics of the image, and can be used for image preprocessing, such as image segmentation, target recognition and transformation detection and the like.

The invention relates to a passive mixer bias circuit, in particular to a passive mixer bias circuit which is used for stabilizing the characteristic of a passive mixer and can follow a threshold voltage of an MOS (metal oxide semiconductor) transistor. The passive mixer bias circuit can stabilize the performances of the conversion gain, the linearity and the noise and the like of the passive mixer in different working environments, and mainly comprises a threshold voltage self-bias reference circuit for generating a current source, and an MOS threshold voltage following device formed by resistors connected in series. The characteristics of the passive mixer (the conversion gain, the linearity and the noise and the like) can be well stabilized by the bias circuit which follows the threshold voltage of the MOS transistor under the condition of different process CORNER, voltage and temperature.

The present invention provides an angular accelerometer signal adaptive denoising method based on wavelet analysis. Compared with a conventional wavelet analysis denoising method, the method of the present invention comprises the steps of calculating the difference quotient of the wavelet energy entropies and the energy entropies of a signal with noise and a noisy signal to determine an optimal wavelet decomposition level of the signal with noise, applying the median filtering to evaluate a noise width and a noise standard deviation, and combining a 3 sigma criterion to set a threshold value to thereby carry out the adaptive filtering. At the aspect of the threshold value selection, the method has better effects by being compared with a conventional heuristic sure criterion, a maximum and minimum value criterion and a penalty criterion. The above methods have better effects on the denoising processing of a periodic signal and a high-dynamic signal of a molecule type liquid-ring angular accelerometer.

Apparatus, systems, and methods to recognize features on bottom surfaces of containers on a container production line, detect defects in the containers, and correlate the defects to specific production equipment of the container production line, based in part on the recognized features. The system includes imaging apparatus, programmable processing devices, and controllers. The methods include imaging techniques and estimation techniques.

The invention discloses a non-local wavelet coefficient contraction-based image denoising method, which mainly solves the problem that image details are lost when denoising is performed by adopting a traditional image denoising method. The non-local wavelet coefficient contraction-based image denoising method comprises the following implementation steps: (1) establishing noise image-contained similarity groups, performing two-dimension wavelet transformation on similarity blocks in the similarity groups, and calculating non-local means of wavelet coefficients of the similarity groups; (2) contracting the wavelet coefficients by using a double-L1 norm model, then, performing wavelet inverse transformation to obtain estimated values of the similarity blocks, and performing integration on the estimated values to obtain primary estimation images; (3) performing residual cover on the primary estimation image, and executing the step (1) and the step (2) to obtain base estimation images; (4) establishing similarity groups of the base estimation images, and further obtaining the noise image-contained similarity groups; (5) performing Wiener collaborative filtering on the noise image-contained similarity groups to obtain denoised images. According to the non-local wavelet coefficient contraction-based image denoising method disclosed by the invention, while noise is smoothened, edge textures of images can be kept better. The non-local wavelet coefficient contraction-based image denoising method can be used for denoising processing of natural images.

An object is to automatically detect the positions of spike noise, create a distribution diagram, and perform pass / fail decisions. The cross-correlation function of the signal waveform from a magnetic head 1 and a reference waveform simulating spike noise is used in extraction of spike noise. The number of peaks in the cross-correlation function exceeding a threshold value is counted, and quantitative evaluation of spike noise is performed.

The invention discloses a latent fingerprint image enhancement method for hyperspectral imaging. The method comprises the following steps: step 1, acquiring a fingerprint endmember spectrum set; step 2, acquiring a background endmember spectrum set based on the fingerprint endmember spectrum set in the step 1 or combination of the fingerprint endmember spectrum set in the step 1 and average spectrum of different background regions marked with a user, and adopting orthogonal subspace iteration for extracting the background endmember spectrum set; step 3, separating fingerprints from backgrounds to realize fingerprint enhancement; and step 4, postprocessing fingerprint features: adopting morphological open-close mixed reconstruction for performing morphology-reserving and noise-inhibiting post-processing on the fingerprint features after enhancement in the step 3. The method is simple and practical, greatly lowers the difficulty of user operation, can smooth noises and fill holes on the premise of keeping enhanced fingerprint information, and is high in fingerprint enhancement reliability, high in efficiency and high in practicality and feasibility.

The present invention discloses a road semi-automatic extraction method and system based on circular templates. The method comprises seven steps of filter pre-processing of road-containing images, acquisition of morphological gradient images, generation of an optimal circular template, acquisition of a reference template gray value, search of to-be-matched templates, encryption of road central points and generation of road center lines. The system comprises an image pro-processing module, a morphological gradient image acquisition module, a circular template generation module, a reference template gray value acquisition module, a matching template searching module and a center line generation module. By combining geometrical characteristics of a road, a multi-direction morphological filter method is adopted to pre-process the images and smooth noises of the road, facilitating follow-up extraction; the circular template is selected at the same time and can well fit the road; the searching speed of the to-be-matched templates is high; and a smooth center line is good in effect by polyline fitting.

A detection system for detecting base contamination at low concentrations in gas, for instance to protect a sensitive process, characterized in that the detection system in constructed to examine multiple amines in gas to produce a reading stoichiometrically related to the proton bonding characteristic of the multiple amines present, the detection system comprising at least two channels through which a gas to be examined passes, an amines remover located in one of the channels, at least one thermal / catalytic converter which discharges NO for each channel, and at least one chemiluminescent NO detector, whereby the total amine concentration is determined from the difference between the detected signals for the channels.

The invention provides an unmanned aerial vehicle positioning method for image registration, which comprises the following steps: (1) preprocessing an image shot by an unmanned aerial vehicle, obtaining the flight height of the unmanned aerial vehicle from a height sensor carried by the unmanned aerial vehicle, and obtaining the flight direction of the unmanned aerial vehicle from a carried heading sensor, obtaining a spatial resolution difference and a direction difference between the unmanned aerial vehicle photographed image and the satellite map according to the marking information of thesatellite map image, and performing rotation transformation and scale transformation on the photographed image to enable the photographed image to have consistent direction and scale with the map image; (2) detecting key points of an image shot by the unmanned aerial vehicle; (3) extracting SIFT features of the key points detected in the images shot by the unmanned aerial vehicle; (4) matching thefeatures of the images shot by the unmanned aerial vehicle with the features of the map images to obtain a corresponding relationship between key point image coordinates in the two images; and (5) estimating the spatial transformation from the unmanned aerial vehicle shot image to the satellite map image, and combining the geographic information of the map to obtain the longitude and latitude ofthe central point of the unmanned aerial vehicle shot image as the current longitude and latitude coordinates of the unmanned aerial vehicle.

A medical image de-noising method based on an optimized nonlocal mean value comprises the following steps of: 1, determining a neighborhood search range and a search parameter for an image to be processed; 2, calculating the similarity between all pixels in the neighborhood range of a current pixel and the current pixel to obtain a neighborhood Gaussian weight matrix; 3, calculating the filtering result of the current pixel according to the neighborhood Gaussian weight matrix, and defining the filtering result in an effective image range; and 4, repeating the steps 2 and 3, completing the traversal of the entire image, and obtaining and outputting a de-noised result image. The image de-noising method establishes an effective calculation boundary condition, and improves the effectiveness of the parameter. A de-noising core algorithm smoothens various artifacts and noise produced by hardware and software of the medical image, and has effective and good detail contrast.

The invention discloses a draining pump which comprises a water pump shell (1), a motor shell (2), an impeller (3), a motor (4) and a water retaining piece (5), wherein the motor (4) drives the impeller (3) to rotate to realize a draining function. The draining pump is characterized in that the impeller (3) comprises a wheel shaft (31) in the middle, upper blades (32) and a flow limiting ring (34), the wheel shaft (31) is connected with a motor shaft (6) of the motor (4), the upper blades (32) are configured on the peripheral side of the wheel shaft (31) around the wheel shaft (31), the flow limiting ring (34) surrounds the upper blades (32), the upper blades (32) transits to a connection portion (321) through a rounded portion (322a) or a chamfered portion (322b) and are connected with the wheel shaft (31) through the connection portion (321), and lower ends of the upper blades (32) are provided with lower blades (33). By the arrangement, flow of the draining pump at the water level limit can be guaranteed, and noises during operation at the low water level can be restricted.

The invention discloses a V-OFDM channel balancing and tracking method. The method comprises the following steps: for a physical layer frame, firstly estimating a channel response initial value H by using a first paragraph of preamble training words; for a first V-OFDM symbol, estimating a pilot frequency value P on a pilot frequency position after a pilot frequency part in the received V-OFDM data symbol is converted onto the frequency domain, balancing the P by using the H, comparing the P with a known pilot frequency sequence Pknown to obtain a compensation value, compensating the channel H to obtain a compensated channel estimation value (the formula is as shown in specification), balancing a data part in the V-OFDM data symbol by using the H, and using the same as a subsequent V-OFDM symbol channel response initial value; and for a second paragraph of preamble training words, estimating a channel response value H2, compensating the channel response value by using the H2 to obtain a compensated channel estimation value (the formula is as shown in specification), and using the compensated channel estimation value as the subsequent V-OFDM symbol channel response initial value. By adoption of the V-OFDM channel balancing and tracking method disclosed by the invention, the channel response estimation value can be continuously updated, so that the balancing operation is faster and more accurate in the case of fast channel change.

The invention discloses a method for comprehensively interpreting infrared detection characteristics of a large-size test piece, which comprises the following steps of: acquiring an infrared thermal image sequence of the large-size test piece from infrared detection, and acquiring an infrared thermal reconstruction image of the large-size test piece from the infrared thermal image sequence; down-sampling the typical type defect infrared thermal reconstruction image in the large-size impact test piece to obtain a down-sampling thermal image containing a relatively low infrared thermal radiation data volume, and executing a multi-target guiding filtering weight acquisition layer step based on the down-sampling thermal image; and based on the multi-target optimal weight ratio parameter, carrying out a multi-target guiding filtering fusion algorithm on the original infrared reconstruction thermal image layer, and finally combining the base layer thermal image and the detail layer thermal image after weighted averaging to obtain a final fusion detection infrared thermal image. Clustering efficiency is improved, the overall detection time of a detection algorithm is shortened, the detection performance of a single infrared thermal image is improved, and the problem that defects of the single detected image are incomplete is solved.

The invention discloses a deep learning method for prostate cancer auxiliary diagnosis. The method comprises: establishing a segmentation network model containing a pre-training model and a chain typeresidual pooling module; carrying out feature extraction and fusion on the input prostate MR images with different scales; and then optimizing the constructed segmentation network by using a random gradient descent algorithm to obtain a trained model, inputting the prostate MR image to be segmented into the trained model, and finally outputting a final segmentation result from a result output bythe model through a conditional random field. The segmentation result of the method can assist doctors in clinical diagnosis and treatment of the prostatic cancer, and the diagnosis accuracy and the working efficiency of the doctors are effectively improved.