56 results about "Spectral transformation" patented technology

The current invention discloses a novel spectral transformation technique for characterizing digitized intensity output patterns from microarrays. This method yields improved sensitivity with reduced false positives and false negatives. Current microarray methods are overly sensitive to the detection of a visible distinction between pixels associated with probes and pixels associated with background. In one embodiment, a technique is disclosed that comprises the steps of: extracting pixels associated with an object of interest and transforming such pixels from an intensity representation to a spectral representation. In some embodiments, the extraction is based on a tessellated logarithmic spiral extraction that may yield a pixel core with a sampling of both foreground and background pixels. This core may then be computationally rescaled by 10×-10,000× to enhance spatial resolution. Once the extracted pixels are represented in the spectral regime, convolution with resolution-enhancement kernels may be used to accentuate morphological features capturing platform specific phenomenology.

The invention discloses a monitoring method of salt ion content in saline soil based on a hyperspectral technology. The method comprises the following main steps: (1) collecting a soil sample in the wild and pre-treating the sample; (2) drying and sifting the soil sample; (3) measuring the total salt and eight salt ions in the soil sample; (4) carrying out hyperspectral measurement on the soil sample to obtain the original spectral reflectance characteristic curve of the soil sample; (5) processing the original spectral value, and carrying out correlation analysis between the spectral value and the measured salt ion content; (6) introducing the construction method of a support vector machine model, and performing model building and precision evaluation on the screened spectral transformation form. The monitoring method is relatively high in the prediction accuracy for each salt ion content in soil, provides a novel approach for the indoor measurement of the salt ion content in saline soil, and provides a novel way for achieving quick and lossless monitoring of the harm types of regional soil salinization.

The invention discloses a qualitative analysis method for improving an identification result on the basis of the near-infrared mode. The qualitative analysis method comprises the steps of (1) acquiring near-infrared spectral data of a sample and determining a modeling set and testing sets; (2) carrying out pretreatment, partial-least-square feature extraction and orthogonal-linear-identification feature extraction on the modeling set and the testing sets in sequence; (3) calculating a spectral transformation matrix between the modeling set and the testing sets by adopting a direct model transferring method, and correcting the remaining testing sets; (4) establishing a quantitative analysis model; and (5) carrying out quantitative identification on the remaining testing sets by utilizing the established quantitative analysis model. The qualitative analysis method disclosed by the invention is established on the basis of near-infrared quantitative analysis, and the orthogonal-linear-identification method which is used in multi-classification and two-classification problems is used in the feature extraction step; in addition, the testing sets can be corrected by the direct model transferring method, so that the model applicability caused by long-time spectral shift of the same instrument can be realized and the result of quantitative identification is improved.

The current invention discloses a novel spectral transformation technique for characterizing digitized intensity output patterns from microarrays. This method yields improved sensitivity with reduced false positives and false negatives. Current microarray methods are overly sensitive to the detection of a visible distinction between pixels associated with probes and pixels associated with background. In one embodiment, a technique is disclosed that comprises the steps of: extracting pixels associated with an object of interest and transforming such pixels from an intensity representation to a spectral representation. In some embodiments, the extraction is based on a tessellated logarithmic spiral extraction that may yield a pixel core with a sampling of both foreground and background pixels. This core may then be computationally rescaled by 10×-10,000× to enhance spatial resolution. Once the extracted pixels are represented in the spectral regime, convolution with resolution-enhancement kernels may be used to accentuate morphological features capturing platform specific phenomenology.

The current invention discloses a novel spectral transformation technique for characterizing digitized intensity output patterns from microarrays. This method yields improved sensitivity with reduced false positives and false negatives. Current microarray methods are overly sensitive to the detection of a visible distinction between pixels associated with probes and pixels associated with background. In one embodiment, a technique is disclosed that comprises the steps of: extracting pixels associated with an object of interest and transforming such pixels from an intensity representation to a spectral representation. In some embodiments, the extraction is based on a tessellated logarithmic spiral extraction that may yield a pixel core with a sampling of both foreground and background pixels. This core may then be computationally rescaled by 10×-10,000× to enhance spatial resolution. Once the extracted pixels are represented in the spectral regime, convolution with resolution-enhancement kernels may be used to accentuate morphological features capturing platform specific phenomenology.

A maximum likelihood spectral transformation (MLST) technique is proposed for rapid speech recognition under mismatched training and testing conditions. Speech feature vectors of real-time utterances are transformed in a linear spectral domain such that a likelihood of the utterances is increased after the transformation. Cepstral vectors are computed from the transformed spectra. The MLST function used for the spectral transformation is configured to handle both convolutional and additive noise. Since the function has small number of parameters to be estimated, only a few utterances are required for accurate adaptation, thus essentially eliminating the need for training speech data. Furthermore, the computation for parameter estimation and spectral transformation can be done efficiently in linear time. Therefore, the techniques of the present invention are well-suited for rapid online adaptation.

A system for processing an image data volume acquired with a medical imaging acquisition device 100 from a body comprising a needle is provided. It has been realized it is advantageous to display 150 a plane intersecting the image data volume showing the needle to a user of the system. This allows better positioning of the needle. The system comprises a needle-plane determination 110 module for determining a plane being parallel to and intersecting a representation in the image data volume of the needle and being parallel to a viewing direction. The needle plane determination module may make use of pixel processing and / or spectral transformation, in particular the Gabor transform.

In a speech recognition system, a method of transforming speech feature vectors associated with speech data provided to the speech recognition system includes the steps of receiving likelihood of utterance information corresponding to a previous feature vector transformation, estimating one or more transformation parameters based, at least in part, on the likelihood of utterance information corresponding to a previous feature vector transformation, and transforming a current feature vector based on maximum likelihood criteria and / or the estimated transformation parameters, the transformation being performed in a linear spectral domain. The step of estimating the one or more transformation parameters includes the step of estimating convolutional noise Niα and additive noise Niβ for each ith component of a speech vector corresponding to the speech data provided to the speech recognition system.

The invention discloses an adjacent harmonic and inter-harmonic separation and measurement method under an IEC (international electrotechnical commission) framework. The method includes the following steps: firstly, performing discrete sampling on a power grid signal; secondly, performing ten-cycle Hanning window addition DFT / FFT (discrete Fourier transform / fast Fourier transform) spectral transformation on a sampling value according to IEC standards so as to obtain a spectrum; thirdly, multiplying the spectrum by rotary phase factors to obtain a new spectrum; fourthly, solving a vector sum of neighboring spectral lines of the new spectrum so as to offset sidelobe interference of rest components on the new spectrum, and solving corresponding frequency, a frequency deviation value and amplitude phase via a spectral line phase cancellation interpolation method; fifthly, rejecting spectrum leakage values of a fundamental component and the rest components at spectral line attention positions on a frequency domain; sixthly, solving a harmonic spectrum value to obtain parameters of h subharmonics, and solving parameters of inter-harmonics adjacent to the h subharmonics. By the method, the problem that the parameters of the two adjacent harmonics and inter-harmonics cannot be measured accurately during limited data asynchronous sampling is solved.

The present invention discloses a downward continuation recursive cosine transform method of a plane geomagnetic anomaly field, and belongs to the field of geomagnetic field data processing. The concrete steps are as follows: preprocessing measurement data of the plane geomagnetic anomaly field; gridding the measurement data of the plane geomagnetic anomaly field; calculating the cosine transformspectrum of the plane geomagnetic anomaly field by means of the recursive cosine transform; setting a small amount epsilon according to the precision requirement of downward continuation, and selecting the value of the damping factor alpha; directly calculating the spectrum of the continuation plane geomagnetic anomaly field through the compensation factors of the cosine transform spectrum on theobservation side and the downward continuation of the plane geomagnetic anomaly field; and carrying out one-dimensional recursive inverse cosine transform in row and in column on the spectrum to obtain a continuation value. According to the method disclosed by the present invention, large-distance downward continuation of the geomagnetic anomaly field can be stably carried out only through one-step spectral transformation and inverse transformation, so that the calculation amount of the downward continuation algorithm is reduced, the recursive algorithm has a small amount of calculation, and the precision of large-distance downward continuation of one-step spectral compensation of the plane geomagnetic anomaly field is improved.

The invention discloses a novel multi-feature fusion electrocardio authentication method. According to the novel multi-feature fusion electrocardio authentication method, multi-feature fusion of a Hilbert spectrum and a decision layer is firstly applied to the authentication method of the electrocardio, the Hilbert spectrum including integrated features of the time domain, the frequency domain and the energy of an electrocardio signal is selected and obtained through EEMD, Hilbert spectral transformation and Hilbert spectrum analysis, a two-dimensional CNN authentication system and a one-dimensional CNN authentication system are used for calculating corresponding matching scores for the electrocardio signal authentication, then the decision layer is merged to make the final authentication decision, and a complete electrocardio certification system with higher robustness and generalization capability is obtained.

The invention discloses a land quality monitoring method based on an aerial hyperspectrum. The method comprises the following steps of step1, collecting aerial hyperspectral data of a land quality monitoring area, and collecting a sample of the land quality monitoring area in the field and carrying out heavy metal element content analysis; step2, pre-processing the aerial hyperspectral data; step3. reconstructing an aerial hyperspectral data spectrum to eliminate radiation distortion of ground and object spectrum caused by various atmospheric components; step4, extracting an aerial hyperspectral image spectrum of a sampling point in aerial hyperspectral remote sensing data; step 5, carrying out spectral transformation and correlation coefficient analysis, obtaining a correlation coefficient between a content and a soil spectral parameter, and finding a sensitive band of a characteristic spectrum; and step6, establishing aerial hyperspectral data to invert a land quality monitoring model, and acquiring a monitored soil nutrient and metal element content data. During application, a wide range of land-based data can be accurately acquired, a workload can be reduced, a land quality monitoring period is shortened, and cost is reduced.

An acoustic signal enhancement method is disclosed. The acoustic signal enhancement method comprises the steps of applying a spectral transformation on a frame derived from an input acoustic signal to generate a spectral representation of the frame, estimating an a posteriori SNR and an a priori SNR of the frame, determining an a priori SNR limit for the frame, limiting the a priori SNR with the a priori SNR limit to generate a final a priori SNR for the frame, determining a spectral gain for the frame according to the a posteriori SNR and the final a priori SNR, and applying the spectral gain on the spectral representation of the frame so as to generate an enhanced spectral representation of the frame. One of the characteristics of the acoustic signal enhancement method is that the a priori SNR limit is a function of frequency.

The invention discloses a real-time monitoring and self-adaptive control system and method for the tail end running state of a numerical control milling machine. The system comprises a vibration sensor, a data collector, a data analysis processor, a data storage server and numerical control equipment. The vibration sensor is installed at the tail end of a spindle or an angle milling head of the numerical control milling machine and used for monitoring vibration signals of the tail end running state of the milling machine in real time, and the data analysis processor conducts spectral transformation analysis processing on collected vibration data, recognizes abnormal information or vibration in the cutting process of the milling machine, stores the processed data in the data storage server,generates a corresponding control instruction according to the identified abnormal information, and sending the control instruction to the numerical control equipment so as to implement adaptive control. The system can identify the abnormal running state of the equipment, early warn potential risks existing in the cutting process of the tail end of the milling machine and immediately send the control instruction, so that the running safety of the equipment is effectively improved, and the enterprise cost is reduced.

Scalable encoding of audio into the core layer based on an ideal noise spectrum determined according to psychoacoustic principles Supports encoding of extension data into the extension layer based on various criteria including offsets from this ideal noise spectrum. Compatible decoding provides multiple decoding resolutions from a single signal. Preferably, sub-band signals resulting from spectral transformation, quadrature mirror filtering, or other conventional processing of the audio input are encoded. A scalable data structure for audio transmission includes a core layer containing the first encoding of the audio signal such that the post-decoding noise is lower than the ideal noise spectrum, and an extension layer containing offset data about the ideal noise spectrum and an extension layer that makes the post-decoding noise lower than the ideal noise spectrum. Decoding data related to encoding of an audio signal with noise below the ideal noise spectrum of the mobile offset data.

The invention discloses a Cr-doped perovskite structure halide near infrared light-emitting material and a preparation method thereof. The chemical composition is AB1-xX3:xCr<3+>, wherein A:B:X is equal to 1:1:3; A is at least one of Cs and Rb; B is at least one of Ag, Na, In, Ga and Pb; X is at least one of Cl and Br; and x is more than 0 and less than 100 mole percent. The invention also discloses the preparation method of the near infrared light-emitting material. The preparation method comprises the following steps: (1) weighing materials: weighing an A-containing compound, a B-containingcompound and a Cr-containing compound separately; and (2) grinding and uniformly mixing the materials and performing high-temperature calcining synthesis by a solid phase method. An emission band of the near infrared light-emitting material is positioned at 800 to 1400 nm and the emission peak is positioned at about 950 to 1050 nm. The preparation method is simple in process, low in raw material price, easy in large-scale technical promotion and is of great importance in research and development of near infrared fluorescent powder with wide spectrum emission. The light-emitting material can beused for manufacturing near infrared light-emitting devices and can be applied in the fields of spectral transformation, radioactive labeling, anti-faking and the like.

The invention discloses a method for constructing a spectral estimation model calibration sample set for total nitrogen in soil. The method comprises the steps of firstly measuring visible-near infrared spectrum data and total nitrogen content data of soil samples; carrying out different spectral transformation on the spectrum data and selecting the calibration sample set from the data after spectral transformation; and finally constructing an estimation model for the total nitrogen content in the soil by using a partial least squares method. Compared with a traditional method for carrying out a Kennard-Stone algorithm to construct the calibration sample set by simply adopting a soil reflection spectrum, the method disclosed by the invention has the advantages that the constructed calibration sample set has spatial representativeness, the prediction ability of the constructed total nitrogen content estimation model is significantly improved and a new ideal is provided for solving the problem that the soil spectrum is comprehensive reflection of soil attributes and the calibration sample set with target component representativeness is difficult to construct.

The invention discloses a method and device for constructing a soil heavy metal environmental risk prediction model in the field of heavy metal prediction. The method comprises the steps of measuringthe content of the heavy metal of the sample soil, and obtaining the heavy metal content data of the sample soil; carrying out spectral measurement and pretreatment on the sample soil to obtain a spectral reflectance curve of the treated sample soil; according to the heavy metal content data of the sample soil and the spectral reflectance curve of the processed sample soil, carrying out spectral transformation and characteristic waveband selection processing, and obtaining a characteristic waveband of the sample soil; constructing a BP-RBF neural network model; inputting the sample soil and the special waveband in the constructed BP-RBF neural network model for training and learning until the BP-RBF neural network model is converged. The construction device comprises a processor and a memory, the BP-RBF neural network is better in accuracy and higher in convergence speed, the precision of the soil heavy metal environment risk prediction is improved, and a new way is provided for the soil heavy metal environment risk prediction.

The invention relates to a laser spot welding monitoring method and a monitoring device, wherein the laser spot welding monitoring method comprises the following steps: in the laser spot welding process, radiation signals of a molten bath are collected; the band-pass filtration is performed for the collected radiation signals to obtain band-pass signals; low-frequency direct-current components and high-frequency noise components in the radiation signals are removed; the fast Fourier transformation and the normalized power spectral transformation are performed for the band-pass signals to obtain molten bath characteristic signals; if the molten bath characteristic signals include spot welding characteristic frequency peaks, spot welding successful signals are output; and otherwise, spot welding unsuccessful signals are output. The laser spot welding monitoring method and the monitoring device can analyze the spot welding quality and determine if the laser power in the spot welding process is proper, have quick analysis speed and low cost, and can be preferably compatible with an existing laser spot welding technology.

The present invention is directed to a method for canceling the Doppler shift, multi-path propagation, delays, and long time selective fading influences of phase and / or amplitude modulated signals carried by RF sub-carriers, without requiring pilot signals. The number of m=2n, over which individual datum is transmitted in each communication channel, is determined. Pairs of Walsh functions are selected from a Walsh-Hadamard matrix such that each function is not selected more than once. Each signal is partitioned to its I factor and Q factor and coding each I and Q factors of the signals by the pair functions. A corresponding phase and amplitude is allocated for each sub-carrier wherein the amplitude and phase values are determined by the data that is transmitted, the digital modulation scheme or inverse spectral transformation, and signs of both corresponding elements taken from the pair Walsh functions. Data is transmitted over the sub-carriers and received and the corresponding phases and amplitudes of all the received signals are obtained using spectral transformation. The received signals are decoded by changing signs of I and Q factors of the received signals in accordance with corresponding elements from the pair functions, used to decode said signals. The I and Q factors of received signals are summed separately, after changing signs of said I and Q factors of received signals, and each of the summations of I and Q factors is divided by m. The signal is reconstructed using said divided I and Q factors, by calculating its phase.

The invention discloses a preparation method of rare-earth aluminoborate fluorescent powder and relates to red fluorescent powder. The invention provides a preparation method of the rare-earth aluminoborate fluorescent powder. A preparation technology is simple; the preparation method is easy to operate and environmentally friendly; the cost is low; and the fluorescent powder is easily excited by ultraviolet. The method comprises the steps of putting lithium carbonate, boric acid, aluminum oxide and europium oxide into an agate mortar for grinding at an element mole ratio of Li:Al:B:Eu of (2-x):1:1:x of a chemical formula Li2 x(AlBO4):xEu<3+>, obtaining a mixture, heating and roasting the obtained mixture in air, and grinding the mixture after cooling to form the rare-earth aluminoborate fluorescent powder, wherein x is greater than or equal to 0 and less than or equal to 0.25. The heating and roasting temperature of the rare-earth aluminoborate fluorescent powder is less than 1000 DEG C; the energy consumption is low; the mass production is facilitated; the fluorescent powder is stable in performance, can be excited by near ultraviolet rays, and emits 570-640nm fluorescent light; and a good spectral transformation material is provided for while LED (light-emitting diode) and crystalline silicon solar batteries.

The invention discloses a method for determining a content of nitrate ions (NO3<->) in secondary saline soil with a near infrared spectrum. The method comprises the following steps: determining the content of NO3<-> of a secondary saline soil sample by using a chemical analysis method; carrying out pretreatment by adopting a soil sample spectrum, establishing a relation model of different variable full-waveband spectrums and content of NO3<-> on the basis of partial least squares (PLSR), and selecting an optimum spectral transformation form; selecting a characteristic waveband of the NO3<-> by adopting an interval partial least squares (iPLS) under the selected spectral transformation, and acquiring a model with the optimum precision, i.e. the relation spectral model of the NO3<-> in the secondary saline soil by virtue of cross verification by utilizing PLSR regression on the basis of the selected characteristic waveband. According to the method, the near infrared spectrum data of the secondary saline soil sample with unknown concentration is determined, the data is preprocessed and substituted in the relation spectral model, and the content of the NO3<-> in the unknown soil sample can be directly obtained.

The invention discloses a method and system for accurately and effectively solving the eigenstate of a large sparse matrix. The method comprises the steps: firstly enabling an original matrix H to beconverted into H' in a mode of spectral transformation, carrying out the psi'through an eigenfunction psi, and precisely giving the eigenvalue and the eigenfunction of the original matrix H through the quick solving of the psi'. According to the method, the solving of the eigenstate is an iterative solving process; according to the method, after an initial parameter is given, self-adjustment is carried out for iterative solving, on one hand, the dependence of the solving process on the initial value is reduced, on the other hand, the convergence rate is increased, the iterative solving efficiency of the matrix is greatly improved, universality is achieved for the bound state and the continuous state in the Hamiltonian quantity, and the method achieves the efficient and quick effects.

The invention relates to a hyperspectral image classification method combining a graph structure and a convolutional neural network. The convolutional neural network can decompose three dimensions into one-dimensional and two-dimensional convolutional neural networks, and the superpixel graph network can be used for subsequent multi-scale two-dimensional processing; the method comprises five steps of hyperspectral image segmentation and spectral feature extraction, multi-scale feature extraction based on a graph network, multi-scale feature extraction based on a convolutional neural network, feature fusion and pixel classification, and loss function and model training. Compared with the prior art, the invention provides a novel multi-scale fusion network and a spectral transformation mechanism, the graphic features and local pixel features based on multi-scale superpixels can be extracted, and the spectral features of graph nodes are extracted in a one-dimensional manner; the method can restrain the noise of an original hyperspectral image, improve the adaptability of the image convolutional neural network to different hyperspectral images, improve the classification precision, automatically extracts the hyperspectral features, and complete the classification. And the classification accuracy reaches over 93%.

The invention relates to the technical field of optical images, in particular to an eye-ground hyperspectral imaging device. The eye-ground hyperspectral imaging device is used for overcoming the problems that in the prior art, spectral transformation in a broad-band continuous wavelength range cannot be achieved and the reliability is poor. The eye-ground hyperspectral imaging device includes a lighting unit, an imaging unit and a controlling unit. The lighting unit includes a light source, a first converging lens, an adjustable concave diffraction grating, a diaphragm, a second collecting lens, a medium altitude reflector and a retinal objective, wherein the first converging lens, the adjustable concave diffraction grating, the diaphragm, the second collecting lens, the medium altitude reflector and the retinal objective are sequentially arranged between the light source and the eye ground in the direction of light path transmission. The imaging unit includes an area-array detector and a controlling computer, a retinal objective, a hollow reflector and an imaging mirror, wherein the retinal objective, the hollow reflector and the imaging mirror are successively arranged between the eye ground and the area-array detector in the direction of the light path transmission; the controlling computer is connected to the area-array detector. The controlling unit includes a controller, a rotating shaft is arranged on the adjustable concave diffraction grating, and the controller is connected to the controlling computer, the light source and the rotating shaft of the adjustable concave diffraction grating separately.

To optimize an imaging range in a depth direction in terms of a relationship with a resolution, an OCT apparatus includes a signal processor that determines a reflected light intensity distribution of an imaging object on the basis of a spectrum of a detected interference light. The signal processor performs spectrum conversion, having a conversion characteristic with which a light source spectrum is converted to a Gaussian distribution curve, on the spectrum of the interference light, and determines the reflected light intensity distribution by Fourier-transforming a spectrum resulting from the spectrum conversion. In the conversion characteristic, the light source spectrum and the Gaussian distribution curve have center wavelengths differing from each other.

A system for processing an image data volume acquired with a medical imaging acquisition device 100 from a body comprising a needle is provided. It has been realized it is advantageous to display 150 a plane intersecting the image data volume showing the needle to a user of the system. This allows better positioning of the needle. The system comprises a needle-plane determination 110 module for determining a plane being parallel to and intersecting a representation in the image data volume of the needle and being parallel to a viewing direction. The needle plane determination module may make use of pixel processing and / or spectral transformation, in particular the Gabor transform.

The invention discloses a fully-enclosed refrigeration compressor valve plate flutter detection method, which comprises the following steps of 1) acquisition of compressor pressure signals, 2) preprocessing of signal data, 3) spectral transformation of the signal data, 4) calculation of the maximum multiplier value of pressure pulsation harmonics and 5) evaluation of valve plate flutter. The fully-enclosed refrigeration compressor valve plate flutter detection method has the beneficial effects that: 1) the detection method can directly detect the fully-enclosed compressor valve plate flutter in actual operation; 2) the detection method does not destroy the structure of the fully-enclosed compressor and does not need to be interposed into the compressor for detection; 3) the detection method is simple to implement, the cost is low, and the detection speed is fast; and 4) the detection method is applicable to online detection of making of the fully-enclosed refrigeration compressor.