699 results about "Nonlinear transformation" patented technology

The invention relates to a method for autonomous flight of a four-rotor unmanned aircraft by using an auto-disturbance-rejection control technique, belonging to the automatic control field of unmanned aircraft. The method comprises the steps of: respectively making differences between an output xld after arranging a transient process of a target value and an output of an extended state observer, and differential of the output xld and the output of the extended state observer respectively, and then carrying out nonlinear conversion on two differences to obtain a nonlinear feedback control law u0; with regard to an auto-disturbance-rejection controller with three attitude angles and vertical displacement, making difference with feedback of the extended state observer to obtain an output as an input of a corresponding channel of a four-rotor system and the extended state observer; and with regard to an auto-disturbance-rejection controller with forward and side displacements, directly using u0 as the input of the corresponding channel of the four-rotor system and the extended state observer, and feeding back the actual value to the extended state observer after the corresponding channel of the four-rotor system responses, so as to form an close-loop auto-disturbance-rejection controller. The method is strong in capacity of resisting disturbance, and the problems of difficulty in modeling of the four-rotor system, environmental diversity in the flight process, and frequent interferences are effectively solved.

This invention relates to a method of performing eye gaze tracking of at least one eye of a user, by determining the position of the center of the eye, said method comprising the steps of:detecting the position of at least three reflections on said eye,transforming said positions to spanning a normalized coordinate system spanning a frame of reference, wherein said transformation is performed based on a bilinear transformation or a non linear transformation e.g. a möbius transformation or a homographic transformation,detecting the position of said center of the eye relative to the position of said reflections and transforming this position to said normalized coordinate system,tracking the eye gaze by tracking the movement of said eye in said normalized coordinate system.Thereby calibration of a camera, such as knowledge of the exact position and zoom level of the camera, is avoided. Further, it is not necessary to know the position of light sources. This results in a much more flexible and user friendly system.

Data cipher processors, advanced encryption standard (AES) cipher system, and AES cipher methods using a masking method perform round operations using a round key, a plain text, a cipher text, and masking data. Some of the round operations are implemented over a composite Galois Field GF(•). Original data and predetermined masking data are processed according to a predetermined rule. Sub-byte transformation operations used in the cipher method and system may include an affine transformation, an inverse affine transformation, an isomorphic transformation, and an inverse isomorphic transformation which are linear transformations, and an inverse transformation that is a non-linear transformation.

Methods and apparatus are provided for spread spectrum communication having spectrum awareness and interference mitigation based on parameterizing an input signal having multiple mixed baseband signals. The method includes non-linearly transforming the input signal, removing carrier effects, removing higher order terms by passing the input signal through a low pass filter to produce a linear combination of data symbols and DC components, solving for the DC components, separating a plurality of mixed baseband signals from the input signal, and coordinate transforming each of the separated signals from polar coordinates to a Cartesian coordinates. A receiver is provided having: a non-linear transform operator receiving a digitized input signal and adapted to produce a linear combination of data symbols, DC components, and carrier effects; a low-pass filter connected to the non-linear transform for removing higher order terms; a separator unit for separating the plurality of mixed baseband signals; and, a coordinate transform for converting the separated signals from polar coordinates to Cartesian coordinates.

The invention discloses a method for effectively recognizing digital modulating signals in non-Gaussian noise. Non-linear transformation is performed on a received signal s(t); the generalized first-order cyclic cumulant and the generalized second-order cyclic cumulant of the received signal s(t) are calculated, and a 2FSK signal is recognized by calculating the characteristic parameters of the received signal s(t) and utilizing a minimum mean square error classifier; the generalized second-order cyclic cumulant of the received signal s(t) is calculated, and by calculating the characteristic parameters of the received signal s(t) and utilizing the minimum mean square error classifier, the number of spectral peaks of a generalized cyclic cumulant magnitude spectrum is detected so that a BPSK signal and an MSK signal can be recognized; the generalized fourth-order cyclic cumulant of the received signal s(t) is calculated, and a QPSK signal, an 8PSK signal and other signals are recognized through the calculated characteristic parameters and the minimum mean square error classifier. The method for effectively recognizing digital modulating signals in non-Gaussian noise solves the problem that signals in Alpha stable distribution noise do not have second or higher order statistics, effectively recognizes the digital modulating signals and can be used for recognizing the modulation mode of the digital modulating signals in the Alpha stable distribution noise.

An adaptive contrast enhancement (ACE) method and apparatus provide a natural enhancement in accordance with the time-varying characteristics of a video sequence. Characteristics of a time varying video sequence are specified and a nonlinear transform over the input video sequence is performed to enhance mainly the contrast of the input. A probability density function (PDF) of a time varying input video sequence is computed and then some predetermined video parameters relating to contrast is extracted from the PDF. Based upon the extracted video parameters, a nonlinear transform function is then constructed and updated as a look up table (LUT), which is synchronized with the associated video picture or field SYNC signal. The transform LUT is then applied to the input video to provide an enhanced video output signal.

The present invention discloses a method and a device for identifying a reticulate pattern face image based on a multi-task convolutional neural network. The method comprises the steps of: collecting reticulate pattern face image and corresponding clear face image pairs, then using the multi-task convolutional neural network to respectively design object functions based on regression and classification, training a face image reticulate pattern removing model, and finally inputting the reticulate pattern face image into the trained reticulate pattern removing model to obtain a face image without reticulate pattern, thereby performing subsequent face image identification tasks. According to the method, a multi-task learning frame is adopted, the task for restoring a reticulate pattern image to a clear image is expressed as two object functions which are assistant with each other, and the convolutional neural network is utilized to learn complicated nonlinear transformation referred therein. The method not only effectively improves convergence rate during model training, but also can greatly improve image restoration effect and generalization ability, thereby greatly improving identification accuracy rate of the reticulate pattern face image.

A “Text Rectifier” provides various techniques for processing selected regions of an image containing text or characters by treating those images as matrices of low-rank textures and using a rank minimization technique that recovers and removes image deformations (e.g., affine and projective transforms as well as general classes of nonlinear transforms) while rectifying the text or characters in the image region. Once distortions have been removed and the text or characters rectified, the resulting text is made available for a variety of uses or further processing such as optical character recognition (OCR). In various embodiments, binarization and / or inversion techniques are applied to the selected image regions during the rank minimization process to both improve text rectification and to present the resulting images of text to an OCR engine in a form that enhances the accuracy of the OCR results.

Processing of masked data using multiple lookup tables (LUTs), or sub-tables, is described. For each input value, an appropriate sub-table provides an output value that is the result of a non-linear transformation (e.g., byte substitution) applied to the input value. An additive mask can be applied to the input data. A transformation can be applied to the masked input data to transform the additive mask into a multiplicative-additive mask. Selected bits of the masked input data and the bits in the additive component of the multiplicative-additive mask can be used in combination to select one of the sub-tables. An entry in the selected sub-table, corresponding to a transformed version of the input data, can then be identified.

A system and methods are provided for extending the frequency bandwidth of a harmonic signal. Harmonic content of a band-limited signal is extended to frequencies outside the signal's passband by performing a non-linear transformation on the complex spectrum of the band-limited signal in the frequency domain. The non-linear transformation may be accomplishes by a linear convolution of the complex spectrum with itself. A system for extending the frequency bandwidth of a harmonic signal includes a signal processor with a forward transform module for transforming a time domain signal into the frequency domain, a non-linear transform module for performing the non-linear transformation on the complex spectrum of the harmonic signal, and a reverse transform module for transforming the extended spectrum of the harmonic signal back into the time domain.

An authentication system using a correlator that correlates an input with a reference wherein at least one of the input and reference comprises a phase volume mask having structures, preferably points, that are each less than about six microns in size and can have an aspect ratio (AR) greater than 1:1 so as to produce a phase encoded random pattern having millions of combinations in a mask that is as small as one square millimeter. The random pattern can be convolved with a second pattern, such as a biometric pattern, to produce a phase convolved mask. The correlator preferably is a nonlinear joint transform correlator that can use “chirp” encoding to permit the input to be located in a different plane than the reference. The correlator optically Fourier transforms images of the reference and input that are thereafter nonlinearly transformed and inverse Fourier transformed by a processor to determine the presence or absence of a correlation spike indicative of authenticity. A spatial light modulator (SLM) can be used as an input or reference and preferably is a liquid crystal panel having pixels or elements whose phase or grey scale intensity can be selectively controlled by a processor. The SLM can be used to display a biometric pattern, preferably scanned in real time from a person, that is correlated against an input or reference that can comprise a label on a card, a tag, or another object.

An image processing method and apparatus performs calculating flash-component image data based on a first image captured without flash emission and a second image captured with flash emission, and generating a finally-adjusted image by using an intensity-adjusted flash-component image generated by executing intensity adjustment on the flash-component image data. The method and apparatus can generate a high quality image in which saturated pixels in a nonlinear transformation image, that is, overexposed highlights are reduced.

An optical touch-sensitive device has the capability to determine touch locations of multiple simultaneous touch events. The touch events disturb optical beams propagating across the touch sensitive surface. With multi-touch events, a single beam can be disturbed by more than one touch event. In one aspect, a non-linear transform is applied to measurements of the optical beams in order to linearize the effects of multiple touch events on a single optical beam. In another aspect, the effect of known touch events (i.e., reference touches) is modeled in advance, and then unknown touch events are determined with respect to the reference touches.

An image processing method and apparatus performs calculating flash-component image data based on a first image captured without flash emission and a second image captured with flash emission, and generating a finally-adjusted image by using an intensity-adjusted flash-component image generated by executing intensity adjustment on the flash-component image data. The method and apparatus can generate a high quality image in which saturated pixels in a nonlinear transformation image, that is, overexposed highlights are reduced.

A method is provided for averaging a sequence of image frames. A noise-reducing filter is applied to the image frames to generate filtered frames. A deconvolution filter is applied to the filtered frames to generate corresponding deconvolved frames. The filtered frames are transformed by an affine transformation to align them, generating aligned frames. The aligned frames are motion corrected by non-linear transformation based on intensity rank matching, generating a sequence of motion-corrected frames. The motion-corrected frames are averaged to generate a resultant frame.

The invention discloses a radio frequency module passive inter-modulation (PIM) interference offset method and device. The device comprises a digital intermediate frequency unit of a radio frequency module, the radio frequency module and an FDD base station system, a PIM component in the radio frequency module is eliminated to a certain extent, and the interference of the PIM on the receiving performance of the radio frequency module is reduced. According to some feasible execution modes, the digital intermediate frequency unit of the radio frequency module comprises an obtaining subunit, a computing subunit and a superposition subunit, wherein the obtaining subunit is used for obtaining digital emitting signals from an emitting channel of the radio frequency module, the computing subunit is used for carrying out nonlinear transformation on the digital emitting signals and generating offset signals for offsetting the PIM component, and the superposition subunit is used for reversely superposing the offset signals to the receiving channel of the radio frequency module so as to offset the PIM component of the received signals.

The invention provides a signal identification and classification method. The method comprises the followings steps of: carrying out noise reduction on initial data containing higher noise by utilizing a wavelet transform method, decomposing signals into high-frequency information and low-frequency information in data analysis, carrying out noise cancelling on the signals by adopting a soft thresholding method and then carrying out signal reconstruction; carrying out further decomposition on the high-frequency part which is not detailedly classified by multiscale analysis while inheriting allthe favorable time-frequency localization advantages of the wavelet transform; analyzing the signals within different frequency bands after multi-layered decomposition by utilizing the wavelet packettransform to extract out characteristic information reflecting a system state; transforming the characteristic vectors of input signals into a high-dimensional characteristic space through non-lineartransform and then solving for an optimal linear classification plane in the high-dimensional characteristic space. The invention overcomes the defects of difficult determination of a network structure, low convergence rate, requirement on large quantities of data samples during training, and the like in neural network learning and enables the neural network learning to be with the characteristics of high precision and strong real time in the aspect of practical application of engineering.

The invention belongs to the field of radar target tracking, and specifically discloses an interactive multi-model radar target tracking method based on predicted value measurement conversion. The interactive multi-model radar target tracking method combines a measurement conversion covariance matrix based on a predicted value with a Kalman filter based on interactive multiple models in a spherical coordinate system, and realizes maneuvering target tracking. Parameters (including estimated prediction distance, azimuth angle, pitch angle and corresponding estimated prediction error covariances) related to a measurement conversion covariance are obtained according to a covariance matrix of an estimated position predicted value in a rectangular coordinate system and position prediction errors obtained by means of the filter, and through nonlinear transformation and a Jacobian transformation matrix. The interactive multi-model radar target tracking method overcomes the inherent defects of a measurement conversion algorithm, and has tracking precision higher than that of an EKF-IMM.

The invention discloses a subsynchronous damping control system, and belongs to the power system stabilization and control technology field. The system includes a rotate speed testing and preposing processor, a combined mode filter, a combined proportion phase shifter, a mode control signal synthesizer and a nonlinear transformer. The system solves the problem of multimode subsynchronous resonance and oscillation of a power system, improves the subsynchronous stability of the power system, and reduces the shafting torsional oscillation fatigue loss of a large-size steam turbo generator.

The invention discloses a method for checking class attendance on the basis of multi-face data acquisition strategies and deep learning. By the aid of the method, the technical problem of low recognition rates of existing methods for checking attendance on the basis of face recognition can be solved. The technical scheme includes that multiple objects are detected and extracted by the aid of AdaBoost algorithms and skin color models. Only a piece of video needs to be shot on every face participating in attendance checking at one step, faces in video sequences are detected and extracted, and face databases can be completely created. The method has the advantages that learning can be carried out on face features in the face databases in different scenes by the aid of simplified LeNet-5 models on the basis of depth convolutional neural network LeNet-5 models by the aid of processes for recognizing the faces on the basis of deep learning, and novel features can be represented by means of multilayer nonlinear transformation; intra-class change of illumination, noise, attitude, expression and the like is removed from the novel features as much as possible, inter-class change generated by identity difference is reserved, and accordingly the face recognition rates of the processes for recognizing the faces in practical complicated scenes can be increased.

The embodiment of the invention discloses a neural network model training method, device and equipment and a medium. The method comprises the steps: acquiring a training sample set comprising trainingsamples and standard label vectors corresponding to the training samples; inputting the training sample into a neural network model comprising a plurality of attention networks; performing nonlineartransformation on the respective output vectors of the attention networks through the neural network model to obtain feature fusion vectors corresponding to the attention networks; and obtaining a neural network model, outputting a prediction label vector according to the feature fusion vector, and adjusting model parameters of the neural network model according to a comparison result of the prediction label vector and a standard label vector until a convergence condition is met, thereby obtaining a target neural network model. The output vectors of all the attention networks are fused in a nonlinear transformation mode, so that the output vectors of all the attention networks are fully interacted, a feature fusion feature vector with more information amount is generated, and the final output representation effect is ensured to be better.

The invention discloses a method for improving image definition in foggy days based on an imaging model. The method for improving the image definition in foggy days based on the imaging model includes the following steps that firstly, an RGB image of original misty days is converted into a YUV color space and three-grade wavelet decomposition is carried out on the luminance Y component; a misty day imaging model is combined on a low-frequency sub-band of a wavelet zone, Gaussian Blur is used for estimating and removing scattered light of media, attenuation factors are adjusted with a method based on local complexity and self-adaption enhancement is performed on an attenuation low-frequency sub-image; a non-linear conversion enhancement method is utilized on a high-frequency sub-band to further enhance high-frequency information; finally, a sharpened colored image is obtained through wavelet inverse conversion and conversion of the color space. According to the method for improving the image definition in foggy days based on the imaging model, the problem of blurring caused by atmospheric scattering can be rapidly and effectively solved, the definition and the contrast ratio of the foggy day image are effectively improved and high real-time performance is achieved.

A method and apparatus for changing a video image having a number of luminance areas to a drawing-style image. An edge boundary between two luminance areas of the video image may be detected and a key signal may be generated therefrom. The key signal may be nonlinear transformed to generate a transformed key signal. Two color signals may be mixed in accordance with the transformed key signal so that the boundary of the video image may be displayable as a line in the drawing-style image.

The present invention discloses a polygon contour similarity detection method. The method comprises: removing an irregular part in figures; establishing a mathematical model of two figures, a complete vector set that describes the figures establishing a feature matrix corresponding to the figures, and calculating an included angle between two adjacent edges; calculating the shortest distance between the two figures; and performing enhanced processing on a calculating result. The polygon contour similarity detection method improves the visual discrimination effect of a machine on figure similarity, and is especially helpful in solving the problem that it is not easy for people to discriminate high-similarity figures; the figure detection effect has relatively high stability and reliability; and the detection time is short, the operation is efficient, and the effect implementation cost is low. According to the polygon contour similarity detection method, only the edges of the figures are inquired, thereby reducing data processing amount. According to the polygon contour similarity detection method, the feature matrix of the figures is constructed, appropriate determination criteria are selected, multi-time enhancement nonlinear transformation is performed on feature matrix elements, and a similarity standard is established by using a multi-value and multi-standard weighted average, so that an algorithm is efficient and has relatively high stability.

The invention provides a device for processing the image of an image sensor and a method thereof. The device mainly includes a gamma correction module, a median filter module and an interpolation module. The method mainly includes the following steps: image data of red, green or blue channels collected by the image sensor is processed by means of nonlinear transforming; according to the magnitude of signal-to-noise ratio of the image data processed by nonlinear transforming, the image data is processed by corresponding median filtering; and the image data of the red, green or blue channels processed by the median filtering is processed by means of interpolatory calculation to obtain the image data of the red, green, blue three channels and the obtained image data is outputted. By using the invention, hardware overhead of the image sensor can be saved and the signal-to-noise ratio of image output of the image sensor can be improved.

The invention relates to a recommendation method and system, and the method comprises the following steps: building a social network of a target user corresponding to an article set consumed by the target user; Establishing a dynamic personal interest model of the target user according to the article set; Constructing a short-term interest model of the social network according to the article set;Constructing a long-term interest model of the social network; Splicing is carried out according to the short-term interest model and the long-term interest model; Calculating the node representationof the target user and the node representation of friends in the social network; Calculating a combined feature weight according to the weight of friends about the target user in the social network; Performing nonlinear transformation on the combined feature weights; Calculating according to the dynamic personal interest model; Obtaining the probability of recommending articles according to the final interest of the user; Calculating a log-likelihood function value according to the probability of the recommended article; According to the technical scheme, the social relation of the user and the dynamic interest and hobby factors of the user can be considered at the same time, so that the recommendation accuracy is improved.

The invention discloses a deep learning-based unusual speech distinguishing method. The method comprises the following steps: input speech is acquired, resampling, pre-emphasis and framing and windowing preprocessing are carried out on the input speech, and preprocessed speech is obtained; a mel-frequency cepstral coefficient (MFCC) characteristic vector is extracted for the preprocessed speech; the speech segments with different frames are regularized to a fixed number of frames, and each speech segment obtains a corresponding mel-frequency cepstral coefficient characteristic vector; a convolutional depth confidence network is built; the mel-frequency cepstral coefficient characteristic vectors are inputted to the convolutional depth confidence network for training, and the states of input speech are classified; and according to a classification result, a hidden Markov model is called for template matching and a speech recognition result is obtained. Multiple nonlinear transform layers of the convolutional depth confidence network are used, the inputted MFCC characteristics are mapped to higher-dimensional space, the hidden Markov model is then used to carry out modeling on different states of speech, and the speech recognition accuracy is improved.

The invention provides a method of hyperchaos pseudorandom sequence generation. (1) select hyperchaos mapping and set initial values thereof x (0) and y(0); (2) carry out iterative computation by input values x(i) and y(i), and obtain output values x(i+1) and y(i+1); (3) undergo nonlinear transformation of x(i+1) and y(i+1), the transformation method is to take the odd bits of x(i+1) and the even bits of y(i+1) to be combined and then take the reverse sequence thereof as the value of the new sequence z(i+1), and at this time the sequence z(i+1) is a real-value sequence; (4) the real-value sequence z is binarized to be a 0-1 sequence, i.e. the pseudorandom sequence output is obtained. The hyperchaos pseudorandom sequence provided in the invention is characterized by strong safety and security, good random properties, simple and rapid generation, and a plurality of autocorrelation properties close to Delta function and cross correlation properties close to zero, being able to be widely applied to the fields like data encryption, secure communication and information safety and so on.

A method of cyphering and / or decyphering, by an integrated circuit, of a digital input code by means of several keys, comprising: dividing the code into several data blocks of same dimensions; and applying to said blocks several turns of a cyphering or decyphering comprising submitting each block to at least one same non-linear transformation and of subsequently combining each block with a different key at each turn, the operands being masked, upon execution of the method, by at least one first random number having the size of the code and all the blocks of which have the same value by combining, by an XOR-type function, the input and output blocks of the non-linear transformation with said random number.

The invention discloses a self-adaptive de-noising and characteristic enhancing method of an SAR (Synthetic Aperture Radar) image, which mainly aims at overcoming the defects that the traditional method has poor de-noising property and can not self-adaptively select parameters in the experiment. The method comprises the following steps: firstly, carrying out logaritmim and then ME-curvelet transformation on an original SAR image; secondly, self-adaptively selecting and optimizing parameters in improved gain functions by adopting an improved PSO (Particle Swarm Optimization) algorithm according to the provided evaluation criterion; and finally, carrying out the nonlinear transformation on ME-curvelet coefficients, ME-curvelet inverse transformation and exponential transformation by adopting the improved gain functions to obtain the final SAR image subjected to the de-noising and characteristic enhancement. By using the method, the noise can be removed while the characteristics are enhanced, the complexity for processing can be reduced and the better de-noising and characteristic enhancing effects of the SAR image can be achieved.