12 results about "Hadamard matrix" patented technology

The application belongs to the technical field of quantum imaging of computational imaging, and specifically discloses a correlation imaging reconstruction algorithm based on a deep convolutional hybrid neural network, which uses a Hadamard matrix gray scale coding sequence to replace a Gaussian matrix speckle sequence of original correlation imaging, and adjusts a digital micromirror device with a fixed sequence of speckle sequences in data acquisition and processing of correlation imaging. Under the same sampling rate, the algorithm can reconstruct a high-quality target object image, and has higher PSNR and SSIM values. Deep learning realizes end-to-end image recovery of an image in the algorithm implementation of correlation imaging, and reduces the steps of image calculation.

The application discloses a kind of ghost imaging reconstruction and classification method and equipment based on hybrid network, belong to the field of computational imaging technology, will be based on Hadamard matrix generation light speckle sequence formed by fixed light speckle sequence;Then, based on this light speckle sequence sampling expanded MNIST data set, generate light intensity sequence, light speckle sequence data;Then, construct and train GRU-CNN network model;Finally, using the model trained to calculate correlation imaging reconstruction and target identification classification.The conventional ghost imaging technique is often faced with the problem of poor image quality under low sampling conditions.In order to solve this problem, the application proposes a GRU-CNN network (GCGI) model combined with deep learning, which can efficiently reconstruct high-quality images and accurately identify and classify targets under very low sampling rates by fusing light intensity sequence and light speckle information, and has strong practical value.

The invention relates to a deep semantic embedding method and device based on a cross-environment separated architecture, and the method comprises the steps: a cloud side generates global semantic embedding through a pre-training language model and a Hadamard matrix based on a class label text, mines a label hierarchical structure through a double-block attention mechanism, and obtains a classification label; generating a high-level semantic hash code containing intra-layer similarity and cross-layer correlation, and constructing an anchor point library; the edge device receives an anchor point library and global semantic embedding, learns a low-level semantic hash code through a discrete cyclic coordinate descent algorithm, and introduces an instance-level fine-grained weight to dynamically distribute a multi-modal data fusion weight; a cloud and an edge device record a corresponding relation of high-layer and low-layer hash codes through a dynamic semantic mapping table, a bit-by-bit calibration algorithm corrects cross-environment deviation, and the edge device only transmits hash code bits different from an anchor point library and compresses semantic information in combination with category-level high-level codes, so that deep semantic embedding with consistent cross-environment semantics is realized. Cloud precision and edge real-time performance are balanced, and efficient retrieval of multi-modal data is supported.

The present application belongs to the technical field of calibration of phased array radar, and relates to a two-stage grouping asynchronous rotating electric vector radar array antenna calibration method, and the specific process is as follows: an attenuator and a phase shifter are configured to each element channel of the array antenna to be measured, and signals received by all channels are synthesized through a power divider; the phased array radar to be measured is equally grouped by channel number in a first stage; a grouping matrix is constructed based on a Hadamard matrix, and the channels grouped in the first stage are divided into rotating groups and fixed groups according to elements in the grouping matrix; a phase shift matrix for phase shift measurement of each row in different grouping matrices is constructed based on a discrete Fourier transform matrix with an order of N, and phase shift is performed; linear equations are constructed according to measurement results and solved, and the ambiguity coefficient, the amplitude of each channel and the phase difference between each channel are obtained, and each field normalization amplitude and phase is solved; amplitude and phase compensation of the channels is completed based on each field normalization amplitude and phase, and finally the amplitude and phase of all channels are consistent.

The invention discloses a spatial channel feature identification method based on compression measurement, and belongs to the technical field of electromagnetic information and multi-antenna wireless communication. According to the method, the random singular value decomposition and compressed sensing theories are combined, programmable array antennas are configured at a transmitting end and a receiving end respectively, the orthogonality of a Hadamard matrix is utilized to realize double-side compressed projection measurement, and a low-dimensional observation matrix is obtained only through a small amount of coding measurement. The main mode of the original space can be deduced by performing singular value decomposition on the matrix, and high-precision estimation of key features such as channel degree of freedom is realized. Compared with a traditional full-amount measurement method, the method has the advantages that the measurement complexity is reduced, the test time and resource consumption are remarkably reduced, and the method has the advantages of being friendly in hardware, high in robustness, high in engineering practicability and the like, and is suitable for scenes such as a large-scale MIMO system, millimeter wave channel modeling and intelligent wireless network optimization.

This application provides a two-dimensional encoding and decoding method for ultrasonic synthetic aperture imaging. The method includes: generating an order Hadamard matrix for an ultrasonic transducer array with n elements, where is a multiple of 8; generating an order serial encoding matrix according to the requirements of axial resolution and near-field dead zone; performing co-encoded transmission and acquisition using n two-dimensional encoding matrices generated by matrix multiplication of the Hadamard matrix and the serial encoding matrix, where ; performing dual decoding of the acquired echo data based on the two-dimensional encoding matrices; and using the decoded data for synthetic aperture imaging. This application can improve the depth and contrast of ultrasonic synthetic aperture imaging.

The invention discloses a cross-border data dynamic management and control method and system based on flow analysis, and relates to the technical field of data transmission, and the method comprises the following steps: encrypting a data stream image according to a Hadamard product matrix and a diagonal algorithm of a matrix vector product, carrying out the first verification of a new data stream according to an encryption verification method, and carrying out the second verification of the new data stream; and performing second verification on the new data stream according to the decryption verification method, and judging whether the new data stream is compliant or not based on the first verification result and the second verification result. According to the method, PPMI weighted co-occurrence and sliding step length minimum Hash position are used, a globally unique and sequenced natural number sequence number is generated for each byte graph, a data stream image and a random Hadamard matrix are subjected to Hadamard product, then diagonalization product is carried out, matrix vector diagonal coefficients are written in an encryption process, secondary confirmation independent of a rule base is provided, and an encryption result is obtained. The sequence number sorting space and the encryption matrix are both updated online along with flow distribution, and the corresponding encryption graph template can be quickly issued to the edge node.

The application discloses a signal fluctuation extraction method combining a Hadamard matrix and mathematical morphology, and comprises the following steps: determining the order m of the Hadamard matrix and the specific number l of rows to be selected according to signal characteristics and sampling accuracy; taking the lth row of the selected Hadamard matrix as a structure element g to perform mathematical morphology operation on an input signal f to obtain a signal fluctuation part and complete the extraction operation. According to the application, a certain row in the Hadamard matrix is selected as the structure element according to signal characteristics, and the order of the matrix is determined; the characteristics that the Hadamard matrix only takes values of 1 and -1 are utilized; the fundamental component in the signal is suppressed through the morphological operation; and the fluctuation part is extracted. The orthogonality of the Hadamard matrix is fully utilized, the Hadamard matrix is used as the structure element for the morphological transformation, the signal can be processed simply and quickly, and the characteristic information is extracted. The application of the mathematical morphology can accurately analyze the signal in the time domain.

The invention relates to a multi-region compressed sensing imaging modulation matrix optimization method. According to the implementation mode, a coding matrix is only arranged at the position with a target to serve as a modulation mask, coefficients of 1 or-1 are given to different coding areas, coefficient sets are linearly independent, the optimized modulation matrix replaces a traditional full-coverage modulation Hadamard matrix to serve as an echo modulation mask, modulated echoes are received by a single-pixel detector, and the modulation efficiency is improved. Two-dimensional information of a target is reconstructed by using a compressed sensing algorithm, depth information is obtained by combining a distance measurement technology, and finally three-dimensional imaging is realized. The problem of spatial modulation redundancy is solved, the coefficient matrix is introduced, one-time imaging of multi-region scenes is supported, high-quality imaging can be completed only by using a low resolution and a small sampling rate, the detection efficiency of complex scenes such as multiple targets and small targets under large-scale terrains is remarkably improved, and the method is suitable for large-scale terrains and large-scale terrains. The method has potential application value in the fields of intelligent driving, high-resolution imaging, multi-target identification and the like.

A weight quantification method and module of a large model, the large model, a device and a storage medium are applied to the field of model quantification, and the method comprises the following steps: obtaining a learnable search parameter having a preset dimension according to position information of model quantification; according to the search parameters and a preset rotation matrix search function capable of being micronized, discrete selection vectors are obtained; optimizing a Walsh-Hadamard matrix candidate library obtained in advance according to the selection vector to obtain a target matrix; performing rotation processing on the corresponding weight matrix based on the target matrix to obtain a rotated weight matrix; and quantizing the weight matrix based on a preset model parameter quantization algorithm to obtain a quantized weight matrix. According to the method, the cache can be successfully quantized to the low bit width, the quantization error is reduced, the quantization precision is improved, and almost all model performances can be reserved. And meanwhile, input and output operations are effectively reduced, the bit width of the Query vector is ensured, and efficient online quantization and calculation are realized.

The application discloses a Hadamard matrix multi-scale sequencing method and system, wherein the method comprises the following steps: initializing parameters, and constructing an M-row and M-column Hadamard matrix; extracting each row or column vector of the Hadamard matrix and then performing normalization, rearranging the elements of each row or column vector to form a 2 n Row 2 n Column square matrix H i ; performing n-layer two-dimensional Haar wavelet decomposition on the square matrix H i , obtaining a coefficient matrix H a , taking the absolute value |H a | of the coefficient matrix H a , summing up |H a |, and recording i into a sequence number value I(i); storing the values of Sa(i) and I(i) into a vector Sa and a vector I respectively; taking i=i+1, and judging whether i is greater than M; arranging the elements in the vector Sa in ascending order to obtain a new sequence R; extracting each row or column vector of the Hadamard matrix after normalization according to the new sequence R, and traversing M times to obtain M two-dimensional 2 n ×2 n square matrices RH i . The application solves the problem that the existing coding cannot simultaneously consider fast imaging and high signal-to-noise ratio imaging.

The invention discloses a unitary space-time modulation and demodulation method and device, electronic equipment and a storage medium, and the method comprises the steps: carrying out the unitary space-time modulation of target data based on a unitary space-time modulation signal constellation set, and obtaining a unitary space-time modulation signal; a first signal in the modulation signal constellation set is generated based on a Hadamard matrix; and converting the unitary space-time modulation signal into an offset quadrature phase shift keying signal, and converting the offset quadrature phase shift keying signal into a continuous phase modulation signal. According to the unitary space-time modulation and demodulation method and device, the electronic equipment and the storage medium, the USTM constellation set head signal is constructed based on the orthogonal characteristic of the Hadamard matrix, the complete constellation set is generated through diagonal matrix design, finally, signal phase continuity is achieved through constant envelope processing, and the peak-to-average ratio can be remarkably reduced.