47 results about "Perfect reconstruction" patented technology

One aspect of the present invention relates to a new alternative to the Bayer pattern for spatial subsampling in color imaging applications. One aspect of the present invention relates to a new design paradigm for spatio-spectral sampling, which is also described. The proposed scheme offers the potential to significantly reduce hardware complexity in a wide variety of applications, while at the same time improving output color image quality. According to another aspect, it is realized that conventional processing techniques are subject to diminishing returns, and with the exception of the most sophisticated processing techniques generate imperfections perceptible to a practiced eye. According to one aspect, a framework for CFA design in presented. In one embodiment the physical characteristics of the CFA are generated so that the spectral radii of luminance and chrominance channels are maximized. In another embodiment, the CFA designed to subject to the conditions of perfect reconstruction. In one aspect, the physical characteristics of CFA design are constrained to require physically realizable CFA(s). Alternatively, certain physical characteristics can be emphasized to generate easier to manufacture CFA(s). According to another aspect, conventional methods and systems are evaluated, confirming shortcomings regarding aliasing in imaging and bandwidth, against which certain embodiments are benchmarked, showing improved performance under at least some embodiments of proposed designs and methods. According to another aspect, a display device CFA is evaluated in terms of throughput of stimuli as limited by aliasing. It is shown the spectral replicas of the chrominance signals induced by existing CFA patterns are centered around frequencies that are not sufficiently far from the DC, consequently overlapping with the luminance signal spectrum and reducing the throughput of the stimuli. By reinterpreting the interactions between the stimuli, display CFA, and CSF in terms of amplitude modulation, an alternative CFA coding scheme that modulates the chrominance signals to a higher frequency relative to common schemes is provided in some embodiments.

Methods and units are shown for supporting a switching from a first coding scheme to a Modified Discrete Cosine Transform (MDCT) based coding scheme calculating a forward or inverse MDCT with a window (h(n)) of a first type for a respective coding frame, which satisfies constraints of perfect reconstruction. To avoid discontinuities during the switching, it is proposed that for a transient frame immediately after a switching, a sequence of windows (h0(n),h1(n),h2(n)) is provided for the forward and the inverse MDCTs. The windows of the window sequence are shorter than windows of the first type. The window sequence splits the spectrum of a respective first coding frame into nearly uncorrelated spectral components when used as basis for forward MDCTs, and the second half of the last window (h2(n)) of the sequence of windows is identical to the second half of a window of the first type.

The invention discloses a semantically-driven camera positioning and map reconstruction method, and belongs to the technical field of computer vision. The method comprises the following steps: firstly, performing semantic segmentation on feature points of a current frame image; matching all the feature points in the current frame and the key frame by adopting a similar matching method according tothe similarity and the semantic category to obtain a matching pair; initializing the posture of the camera through all matching in the current frame and the key frame; updating the feature point matching pairs by adopting a three-dimensional projection method in combination with semantic judgment; updating all the feature point matching pairs by utilizing attitude minimization; and finally, constructing a three-dimensional map by utilizing the camera attitude. The invention further discloses a semantic-driven camera positioning and map reconstruction system. According to the technical schemeprovided by the invention, not only is a plurality of processing performed in a camera positioning stage, but also point cloud constraint is performed in a reconstruction stage, so that semantic segmentation is more tightly combined with a camera positioning and reconstruction system, and a more accurate positioning result and a more perfect reconstruction effect are obtained.

This invention introduces a class of multi-band linear phase lapped biorthogonal transforms with fast, VLSI-friendly implementations via lifting steps called the LiftLT. The transform is based on a lattice structure which robustly enforces both linear phase and perfect reconstruction properties. The lattice coefficients are parameterized as a series of lifting steps, providing fast, efficient in-place computation of the transform coefficients as well as the ability to map integers to integers. Our main motivation of the new transform is its application in image and video coding. Comparing to the popular 8.times.8 DCT, the 8.times.16 LiftLT only requires 1 more multiplication, 22 more additions, and 6 more shifting operations. However, image coding examples show that the LiftLT is far superior to the DCT in both objective and subjective coding performance. Thanks to properly designed overlapping basis functions, the LiftLT can completely eliminate annoying blocking artifacts. In fact, the novel LiftLT's coding performance consistently surpasses that of the much more complex 9 / 7-tap biorthogonal wavelet with floating-point coefficients. More importantly, our transform's block-based nature facilitates one-pass sequential block coding, region-of-interest coding / decoding as well as parallel processing.

A subband audio coder employs perfect / non-perfect reconstruction filters, predictive / non-predictive subband encoding, transient analysis, and psycho-acoustic / minimum mean-square-error (mmse) bit allocation over time, frequency and the multiple audio channels to encode / decode a data stream to generate high fidelity reconstructed audio. The audio coder windows the multi-channel audio signal such that the frame size, i.e. number of bytes, is constrained to lie in a desired range, and formats the encoded data so that the individual subframes can be played back as they are received thereby reducing latency. Furthermore, the audio coder processes the baseband portion (0-24 kHz) of the audio band-width for sampling frequencies of 48 kHz and higher with the same encoding / decoding algorithm so that audio coder architecture is future compatible.

This invention introduces a class of multi-band linear phase lapped biorthogonal transforms with fast, VLSI-friendly implementations via lifting steps called the LiftLT. The transform is based on a lattice structure which robustly enforces both linear phase and perfect reconstruction properties. The lattice coefficients are parameterized as a series of lifting steps, providing fast, efficient in-place computation of the transform coefficients as well as the ability to map integers to integers. Our main motivation of the new transform is its application in image and video coding. Comparing to the popular 8×8 DCT, the 8×16 LiftLT only requires 1 more multiplication, 22 more additions, and 6 more shifting operations. However, image coding examples show that the LiftLT is far superior to the DCT in both objective and subjective coding performance. Thanks to properly designed overlapping basis functions, the LiftLT can completely eliminate annoying blocking artifacts. In fact, the novel LiftLT's coding performance consistently surpasses that of the much more complex 9 / 7-tap biorthogonal wavelet with floating-point coefficients. More importantly, our transform's block-based nature facilitates one-pass sequential block coding, region-of-interest coding / decoding as well as parallel processing.

The invention relates to a differential protection method using single-ended transient signal high-frequency component throughputs of a transmission line, belonging to a single-ended transient protection scheme for a power transmission line. When the line is in short circuit, a large number of high-frequency signals are generated due to a fault and are propagated to the two ends of the line; and as a bus has ground distributed capacitance, the high-frequency signals are remarkably reduced when the high-frequency signals pass through the bus, and high-frequency component values detected at a proximal fault side of the bus are always greater than those detected at a distal fault side of the bus. According to the differential protection method, fault signals of the inner side and the outer side of the bus at a line protection mounting position are acquired by using the above-mentioned principle, fault transient signals are subjected to lossless segmentation by using a perfect reconstruction filter bank, then, the difference value of certain throughputs of subband components of the inner side and the outer side of the bus is calculated, and whether the fault is an internal fault or an external fault is judged according to a majority voting mechanism by using the difference value; and signals for constructing criteria are a plurality of subband high-frequency signals covering the full high-frequency band, and subband frequencies can be adaptively adjusted along with the change of a sampling frequency. The differential protection method has the advantages that the scheme is simple and reliable, the line of total length can be protected, the requirements on the distributed capacitance of the bus are low, and the fault tolerance and the reliability are enhanced from principle and hardware implementation manners.

Many methods for reversible watermarking (embedding schemes that allow perfect reconstruction of the original host signal) are highly fragile in the sense that the slightest modification of watermarked content prohibits the recovery of both the original signal as well as the embedded auxiliary data. In order to obtain robustness against transmission or channel errors, the embedding method according to the invention accommodates error correction data in a portion of the data embedding capacity. In an advantageous embodiment, the host signal (36) is segmented in segments, and error correction data (p(n)) for a segment (S(n)) is accommodated in data (37) being embedded in a subsequent segment (S(n+1)) along with restoration data (r(n)) for reconstructing the host signal. The remaining portion of the embedding capacity is used for payload (w).

A filterbank system is provided for reordering subbands of a wideband digital signal. The filterbank system extracts the subbands from a wideband digital input signal into an even channel number group of even channel number subbands and an odd channel number group of odd channel number subbands. The separation of even and odd channel number subbands provides an even channel grouping with guard bands between the even channel number subbands and an odd channel number grouping with guard bands between the odd channel number subbands. The filterbank system then reorders the even channel number subbands and the odd channel number subbands. The reordered subbands are then combined by combining the reordered even channel number subbands and odd channel number subbands into a wideband digital output signal.

The invention relates to a satellite-borne reconfigurable broadband digital channelizing switching method, belonging to the technical field of satellite communication. The method comprises the following steps of: carrying out secondary sampling to uplink signals of the satellite so as to obtain digital signals; constructing perfect reconfigurable prototype filtering module and obtain a complex-exponential modulation perfect reconfigurable filter group module by complex-exponential modulation; carrying out filtering to the secondary sampling signals so as to obtain baseband signals of subchannels; carrying switching to the separated baseband signals of the subchannels according to switching information set by the satellite so as to obtain switched data; carrying combination to the switched data so as to obtain the downlink signals of the satellite. The method is not independent physical layer concrete setting, is not relevant to the wave form of the signals to be processed, and has good adaptability; the method is suitable for uniform bandwidth channel switching scenes and is also suitable for non-uniform bandwidth channel switching scenes; all the control is realized by control signaling with flexible reconfigurability; and the method supports functions of broadcast and multicast simultaneously.

A pair of M-path perfect reconstruction filter banks, one for analysis and one for synthesis avoid aliasing of channel filter band-edges by operating at an output sample rate of 2-samples per channel bandwidth. The non-maximally decimated analysis filter performs an M / 2-to-1 down sample operation to obtain the 2-samples per channel output rate. The corresponding synthesis bank accepts signal samples at 2-samples per symbol and up-samples 1-to-M / 2 to obtain the original and desired output sample rate matched to the input sample rate. Changes in composite system bandwidth are achieved by enabling or disabling the connection between the analysis filter bank and the corresponding synthesis filter bank.

A system and method is provided for efficient analog reconstruction of a digital signal to an intermediate frequency (IF). The digital signal can be a complex baseband digital signal that is sourced from a near perfect reconstruction filterbank (NPRFB). The output of the NPRFB, after some manipulation in the digital domain, is converted to the analog domain at IF. The IF analog signal can then be modulated with a high frequency RF signal or further mixed with a RF transmission frequency signal for RF transmission.

The invention discloses a design method of a two-dimensional double prototype completely-oversampled DFT (discrete Fourier transform) modulated filter bank. The design of the completely-oversampled DFT modulated filter bank is realized under the condition of near perfect reconstruction. The design problem of two prototype filters is resolved into an unconstrained optimization problem, wherein an objective function is overall distortion of the filter bank, namely, a weighted sum of transmission distortion, aliasing distortion and stopband energy of the prototype filters, and the optimization problem is solved through a double-iteration mechanism according to gradient vectors of the objective function. The computation complexity is remarkably reduced according to equivalent conditions of matrix inversion and a fast algorithm of Toeplitz matrix inversion in single-step iteration. With the adoption of the method, the filter bank with better overall performance can be obtained, the computation complexity is greatly reduced, and the large-scale two-dimensional filter banks can be rapidly designed.

The invention relates to the audio signal processing technology and discloses a method and equipment for processing digital audio in variable speed. In the invention, a pair of perfect reconstructing window functions WL and WR with amplitude attenuation and increase characteristics is used to act on original digital audio according to different delays to obtain a pair of windowing data; and an audio waveform is reconstructed by using the windowing data to obtain audio after variable-speed process. The defection on a fundamental tone period and the relativity of the audio and the time-frequency conversion are avoided, so that the calculation amount is extremely small. In addition, the playing time of the contents is prolonged or shortened by using the compaction and the introduction of theself waveform of an audio signal without changing the audio waveform, so that the original tone quality can be better maintained.

The invention discloses a voice conversion method. Firstly, a feature extraction algorithm is used for reducing dimensionality of a model parameter; Mel frequency is set in advance to be used as sampling frequency, and the STRAIGHT spectrum of a voice signal is sampled, so that a Mel frequency logarithmic spectrum is obtained; a standard deviation is calculated by using MFLS; near perfect reconstruction is carried out on the STRAIGHT spectrum through a Gaussian mixture model by using the Mel fixed frequency, the MFLS and the standard deviation, so that feature parameters are obtained; secondly, the method disclosed by the invention is different from a conventional method for estimating a statistic model from training data; by establishing a condition histogram and approximately evaluating conditional distribution of target voices, voice conversion is realized; the characteristics of limited own resources of an embedded system, a mobile device and the like are fully combined, so that high efficiency, short consuming time, simple calculation and low energy consumption are realized.

The invention discloses an optimum design method for an M channel oversampling image filter set based on convex optimization. By means of a two-step method, target optimization is performed, the spectral selectivity and perfect reconstruction conditions of the image filter set are comprehensively considered, and therefore it is ensured that the whole performance of the image filter set is improved under the small reconstruction error condition. The simple and effective solution is provided for lowering the reconstruction error of the image filter set and achieving signal restoration and reconstruction.

A resource reuse perfect reconstruction sub-band analyzer comprises an interpolation module, a 2m dot symbol adjusting module, an FIFO module, an N-way multiplication module, an N-way addition module, a complex multiplication module, a sub-band filter coefficient table, a complex multiplication coefficient table, a 2m dot FFT module and a perfect filter factor adjusting module. The resource reuse perfect reconstruction sub-band analyzer can conduct 2m times reuse on hardware resources, the expenditure of hardware resources is lowered, the structure is simple and universal, parameter configuration is flexible, the technological indexes such as the number of ways, the speed and the bandwidth for sub-band analysis can be rapidly changed, and the resource reuse perfect reconstruction sub-band analyzer is suitable for various bandwidths and speeds and capable of being widely applied to mobile satellite communication systems, software-based processing transponder satellite systems, ground face mobile communication systems and various data collecting systems.

The present invention provides a resource reuse perfect reconstruction sub-band synthesis processing method. The method comprises the following steps of 1 carrying out the perfect filtering factor adjustment processing on an input signal; 2 carrying out the IFFT and the coefficient correction; 3 adopting N FIFO which are in head-to-tail ligation to carry out the data cache and output; 4 adopting a reconstruction filtering coefficient existing in a matrix to carry out the filtering process on the output data of the FIFO; 5 carrying out the extraction and symbol adjustment on a filtering output signal. The method can realize the 2m times reuse of the resources, and a resource reuse multiple is directly proportional with a sub-band number, namely, the larger is the sub-band number, the more is the resource reuse times, and the more obvious is the advantages of the present invention. A sub-band synthesizer realized by the method enables the hardware resource cost to be reduced substantially, is flexible in parameter configuration and wide in application range, and is suitable for the systems of different bandwidths and rates to carry out the sub-band synthesis.

The invention relates to a cone beam CT circular-linear track back projection filtering reconstruction method. M-lines are selected according to the imaging geometry characteristics of a circular-linear scanning track; a back projection integral interval is determined by calculating the endpoint coordinates of an R-line passing through the reconstruction points on the M-lines; and object reconstruction is implemented through Hilbert transform. The problem of cone angle effect during big-cone-angle scanning of a single circular track can be solved effectively. A linear scanning track is added on the basis of circular track scanning, so the conditions of perfect reconstruction are satisfied, and complete reconstruction projection data is obtained. The method can be better applied to actual scanning, effectively improves the quality of an image reconstructed at a big cone angle, and is of important practical significance to the expansion of the application range of cone beam CT scanning.

The invention relates to an ultra-wideband pulse radio preprocessor system capable of being used for interference elimination. The system comprises a subband signal division unit and a subband signalrecombination unit, the subband signal division unit is used for dividing input ultra-wideband pulse radio signals into M subband signals, the subband signal recombination unit is used for recombiningthe M subband signals output by the subband division unit, wherein M is an integer greater than or equal to 1. The system provided by the invention has no need of using a digital-to-analog converterwith ultrahigh conversion rate, and overcomes obstacles of difficulties in digital-to-analog conversion caused by ultrahigh bandwidth in ultra-wideband pulse radio signal digital processing or interference processing; and, in the system provided by the invention, only the subband signals suffered from interference need to be processed by digital-to-analog conversion, and subjected to frequency shift to a low frequency region, thus, sampling rate can be reduced greatly; moreover, the system provided by the invention can realize perfect recombination of input signals by itself, and has excellentrobustness in case of that each branch has defects or is mismatched.

The invention discloses a design method of a near-perfect reconstruction mono-prototype DFT modulated filter bank. According to the design method, a filter bank with good overall performance is obtained by relieving the transmission distortion of the filter bank and stopband attenuation of a prototype filter. The design problem of the prototype filter is resolved into an optimization problem with constraint, stopband energy of the prototype filter serves as an objective function, transmission distortion of the filter bank serves as the constraint, and the prototype filter is solved with the iterative method. Reconstruction errors of the filter bank reach a low level through many times of iteration. The design method provides a simple and efficient solution for the near-perfect reconstruction mono-prototype DFT modulated filter bank.

The invention discloses a wavelet tight frame adaptive sparse three-dimensional seismic data reconstruction method, which is applied to the field of seismic data processing, and aims at the limitationof traditional fixed base sparse transformation and the problem that an existing adaptive sparse transformation method is insufficient in stability. Through implementing the wavelet tight frame adaptive sparse three-dimensional seismic data reconstruction method, the stability of dictionary learning is improved by using a wavelet tight frame, the optimal sparse representation of data is obtainedthrough adopting a stable dictionary learning method, and the reconstruction of the three-dimensional seismic data of a complex structure is realized and the noise reduction effect is good by using the perfect reconstruction characteristic of the wavelet tight frame; and by using the seismic data reconstructed by adopting the wavelet tight frame adaptive sparse three-dimensional seismic data reconstruction method, the reliability of the underground information obtained by three-dimensional seismic exploration is improved, more accurate underground information is provided for the next production work, and the purposes of improving the production quality and reducing the economic cost are achieved.

There is disclosed in embodiments methods for halftone dot encoding for engine-dependent-rendering. The methods first carefully design an asymmetric halftone dot for a specific printer and then utilizes a halftone dot encoding algorithm, which encodes the asymmetric halftone dot into a symmetric format. The asymmetric halftone dot design is satisfies the requirements of raster output scanner electronics, as well as several engine dependent constraints. These include: rotation insensitivity, minimum dot size, minimum hole size, minimum appendage, and minimum cavity. The symmetric dot encoding allows perfect reconstruction of the binaries that may have been generated using the original asymmetric dot. This reconstruction is achieved using imbedded binary resolution conversion. In this way, this resolution conversion is acting as a decoder.

A system and method is provided for efficient analog reconstruction of a digital signal to an intermediate frequency (IF). The digital signal can be a complex baseband digital signal that is sourced from a near perfect reconstruction filterbank (NPRFB). The output of the NPRFB, after some manipulation in the digital domain, is converted to the analog domain at IF. The IF analog signal can then be modulated with a high frequency RF signal or further mixed with a RF transmission frequency signal for RF transmission.

The invention provides a DeepFake defense method and system based on visual adversarial reconstruction. The method comprises a two-stage method. In a first stage, real face data are converted into potential codes, and real images can be realistically reconstructed through a generator. The problem is regarded as a reverse problem of the GAN, an encoder is trained to generate potential embedding, the potential embedding is used as initialization, and then fine adjustment is carried out on the potential embedding. In a second stage, a search is performed in the neighborhood of the potential embedding obtained in the first stage to obtain an optimal embedding, which can result in a perfect reconstruction and invalidate the DeepFake. In the process, the potential embedding is optimized by using gradient information from the target DeepFake model and is limited in a small modification range so as to meet the visual similarity requirement.

The invention discloses a TIADC system sampling moment mismatch digital background calibration method with low hardware overhead. On the basis of Lagrange interpolation, a reconstruction filter is realized to compensate the mismatch at the sampling moment. By simplifying the polynomial expression of the interpolation coefficient, compared with other perfect reconstruction methods, the architectureprovided by the invention has the advantages that about 41% of multipliers are saved, and the number of adders is also greatly reduced. Simulation verification is carried out in a four-channel 800MSPS 12-bit TIADC system, and the result shows that the TIADC system can obtain at least 72dB SNDR within the bandwidth of 0.4 fs input signals. Due to the full-digital realization characteristic of thecalibration algorithm, the method is very suitable for being applied to FPGA or DSP equipment to realize data post-processing. Meanwhile, in a high-speed multi-channel TIADC system, the number of multipliers and adders which are greatly reduced is more easily integrated in a chip, and conversion between different advanced process nodes is more flexible.

The invention discloses a distributed compact sensing matrix tracking method based on a cross array radar system, comprising the following steps: (1) obtaining compact sensing matrixes psiX and psiY according to sensing matrixes phiX and phiY corresponding to two arrays in a cross array radar; (2) carrying out processing by an orthogonal matching tracking algorithm to obtain initial estimates S(^)<ini><X> and S(^)<ini><Y> of a correct subspace; (3) forming two candidate column sets f(^)<X> and f(^)<Y> according to all the columns in a plurality of similar column groups corresponding to the initial estimates S(^)<ini><X> and S(^)<ini><Y>, performing union operation on the two candidate column sets to obtain the latest candidate column set f(^), performing permutation and combination on the columns to find a column combination minimizing the residual sum and obtain a correct support set; and (4) obtaining corresponding azimuth angle and pitching angle according to the correct support set. By constructing low-coherence compact sensing matrixes, the problem that a sparse vector cannot be perfectly reconstructed with large probability due to the fact that the original perception matrix has high coherence as the radar resolution increases is overcome, and the result is more accurate through fusion of the information of two columns.

The invention discloses a design method for approximately completely reconstructing a non-uniform cosine modulation filter bank. The non-uniform filter bank reconstruction error obtained by directly optimizing the performance of the non-uniform filter bank is smaller and the overall performance is more excellent. . Simultaneously the present invention is by summarizing the design problem of the non-uniform filter bank into an unconstrained optimization problem about the prototype filter, wherein the objective function is the weighted sum of the transfer distortion of the non-uniform filter bank and the stopband energy of the prototype filter, and finally utilizes the linear An iterative algorithm is used to solve the optimization problem, which significantly reduces the design cost of the non-uniform filter bank. Therefore, the present invention provides a simple and efficient solution for reducing design complexity and realizing accurate signal reconstruction.

The invention relates to a method for down-sampling and reconstruction of signals with limited innovation rate based on a linear regular domain, and belongs to the technical field of digital signal processing. The innovation of this method is: based on the characterization of the signal in the linear regular domain, by designing a new type of sampling kernel with tight support, the finite innovation rate signal can be down-sampled, and then the signal reconstruction can be completed through high-resolution spectral estimation. . Compared with the prior art, according to the finite innovation rate of the signal and the four free parameters of the linear canonical transformation, the method designs different new tight-supported sampling kernels, which can realize the signal sampling with down-sampling data far lower than the traditional Nyquist sampling rate. Perfect reconstruction breaks the requirements and limitations of Nyquist sampling theory on signal band limit and sampling rate. It is of great significance to reduce the requirements of digital-to-analog conversion equipment, reduce storage space and computing energy consumption, and further promote the development of the information field.