Patents
Literature

51results about How to "Improve detection performance" patented technology

Spectrum sensing algorithm based on variational mode decomposition

The invention discloses a spectrum sensing algorithm based on variational mode decomposition, and solves the problem that a signal at a receiving end is interfered by a noise background in the prior art. The method comprises the following steps of: 1, generating a simulation signal BPSK; 2, taking the mean value of the instantaneous frequencies of the components as a decomposition parameter, determining an optimal mode decomposition number K according to a change curve of the decomposition parameter, and setting an appropriate penalty factor alpha; 3, performing variational mode decompositionon the BPSK signal according to the obtained decomposition parameters; 4, respectively carrying out power spectrum estimation on the obtained L intrinsic modes; 5, taking the ratio of the sum of the spectral line intensities of the intrinsic modes to the sum of intensities of all the intrinsic modes as a test statistic; and 6, comparing the obtained test statistics with a judgment threshold, and making a final judgment. According to the method, the equivalent band-pass filtering characteristic and the Wiener filtering structure of variational mode decomposition are utilized, frequency bands are better divided, and denoising processing is better carried out; and compared with a traditional power spectrum segmented cancellation algorithm, the detection performance can be remarkably improved.
Owner:XIDIAN UNIV

Cooperative spectrum sensing method based on double-threshold energy detection

InactiveCN104601263AImprove detection performance
The invention discloses a cooperative spectrum sensing method based on double-threshold energy detection. The method includes the steps of performing double-threshold local determination on data perceived by a perception user locally, determining a frequency band to be occupied if the data is greater than the maximum and sending Presence to a data center, determining the frequency band to be idle if the data is smaller than the minimum and sending Absence to the data center, and packing and then sending statistics and the signal to noise ratio to the data center if the data is smaller than the maximum and greater than the minimum; storing the received statistics and the received signal to noise ratio in the data center, directly determining the frequency band to be occupied if the data center receives any one of Presence, reading the statistics and the signal to noise ratio of every perception user if the data center does not receive any Presence, utilizing the maximal-ratio combining principle to obtain the weighting factor of every statistic for further obtaining statistics after combination, and performing single-threshold cooperative spectrum sensing on the statistics after combination. The cooperative spectrum sensing method based double-threshold energy detection can detect whether different frequency bands are idle or not, is applicable to different noise environments, dynamically sets a soft decision threshold and improves the detection performance of a system.
Owner:NANJING UNIV OF POSTS & TELECOMM

Full coherent full polarization MIMO radar four-channel integrated target detecting method

InactiveCN106597381ALarge amount of informationImprove detection performance
The invention discloses a full coherent full polarization MIMO radar four-channel integrated target detecting method, which is used to solve a problem of a conventional non-coherent accumulation detector that detection performance is lowered, when signal-to-noise ratios of echo signals of four polarization channels are different. The full coherent full polarization MIMO radar four-channel integrated target detecting method comprises steps that 1) a full coherent full polarization MIMO radar is used to transmit signals, the waveforms of which are orthogonal to each other, by orthogonal polarization channels; 2) the full coherent full polarization MIMO radar is used to receive the echo signals by the orthogonal polarization channels; 3) the four-way pulse compression of the echo signals received by the orthogonal polarization channels is carried out, and the echo signals of the four polarization channels are acquired; 4) the signal-to-noise ratio weighting integrated detection of the echo signals of the four polarization channels is carried out. Compared with the prior art, the full coherent full polarization MIMO radar four-channel integrated target detecting method has constant false alarm performance, and the better detection performance is also provided, and therefore the full coherent full polarization MIMO radar four-channel integrated target detecting method can be used to realize the target detection of the full coherent full polarization MIMO radar under a noise background.
Owner:XIDIAN UNIV

Energy detection method based on double-threshold power spectrum double-average ratio

InactiveCN110601779AImprove detection performanceImprove performance and robustness
The invention discloses an energy detection method based on a double-threshold power spectrum double-average ratio, and relates to the technical field of cognitive radio, and the method comprises thesteps: 1, calculating double thresholds and a detection statistics TDTED under double-threshold energy detection, judging whether the detection statistics TDTED is greater than a lower threshold gamma0 and less than an upper threshold gamma 1 or not, and if not, executing the step 2; if so, executing the step 3; 2, judging whether the detection statistics TDTED is greater than or equal to an upper threshold gamma 1 or not, if so, determining that the PU signal exists, and if not, determining that the detection statistics TDTED is less than a lower threshold gamma 0, and determining that the PU signal does not exist; 3, calculating the maximum value and the minimum value of the power spectrum, the test statistics TDPSRA and the threshold value gamma DPSRA by adopting a power spectrum double average ratio algorithm; and 4, judging whether the test statistics TDPSRA is greater than or equal to a threshold gamma DPSRA, if so, determining that the PU signal exists, and if not, determiningthat the PU signal does not exist. According to the invention, another detection algorithm for processing is added between double thresholds, so that the processing efficiency of the algorithm is improved, and the detection performance of energy detection can be remarkably improved.
Owner:CHANGCHUN UNIV OF SCI & TECH

Pedestrian detection network, model training method, detection method, medium and equipment

The invention discloses a pedestrian detection network, which takes VGG16 as a basic network and comprises five characteristic layers, namely fc6, fc7, conv6 _ 2, conv7 _ 2 and conv8 _ 2. The invention also discloses a pedestrian detection network model training method. The method comprises the following steps: calculating an expected value E of the pedestrian frame width-to-height ratio; extracting features to obtain a confidence map and a coordinate bias map; making a dense frame generation strategy; preprocessing the training data file, and training a pedestrian detection network to obtaina pedestrian detection network model. The invention also discloses a pedestrian detection method, which comprises the following steps of: setting a detection threshold value, loading the pedestrian detection network model and initializing network parameters; sending the picture into a model to obtain a detection result; filtering all the detection frames through non-maximum suppression to remove the frames detected repeatedly, and finally storing the result in a file. The method is based on the dense frame generation network, effectively overcomes the defects of high leak detection rate and the like in a pedestrian dense scene in the prior art, and improves the detection performance of the frame in the pedestrian dense scene.
Owner:SOUTH CHINA UNIV OF TECH +1

End-to-end multi-view three-dimensional human body posture estimation method and system and storage medium

ActiveCN112560757AImprove detection performanceImprove continuity and anti-jitter
The invention relates to an end-to-end multi-view three-dimensional human body posture estimation method and system and a storage medium, and the method comprises the steps: loading a pre-trained two-dimensional human body posture estimation network, and enabling a picture of each current view to serve as the input of the network; generating a thermodynamic diagram through a two-dimensional humanbody posture estimation network, and taking the thermodynamic diagram as the input of an LSTM thermodynamic diagram time sequence information extraction network; inputting the thermodynamic diagram into an LSTM initialization thermodynamic diagram time sequence information extraction network and an LSTM thermodynamic diagram time sequence information extraction network according to the value of the time sequence step length T to obtain a cell state and a hidden state; feeding the obtained hidden state into a decoder network to obtain a decoded thermodynamic diagram; fusing the thermodynamic diagram and the decoded thermodynamic diagram to obtain a thermodynamic diagram Ht (p) fused with time and space information; sending the thermodynamic diagram Ht (p) into a soft-argmax linear algebraictriangulation network to obtain a 2D point position; and solving an overdetermined equation on the homogeneous three-dimensional coordinate vector, and adopting a differentiable DLT-SII algorithm toobtain a final three-dimensional human body posture estimation point.
Owner:UNIVERSITY OF CHINESE ACADEMY OF SCIENCES
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products