340 results about "Belief propagation" patented technology

The invention discloses a channel-polarization-based encoder for a staggered structure duplication code, and an encoding method and decoding methods thereof. The encoder consists of a duplication bit buffer with a storage capacity of L bits, a bit position mapper with a length N and a channel polarization device with the length N which are connected in sequence. The encoding method based on the encoder comprises the following steps of: embedding duplicated encoding into a channel polarization process, and introducing a duplicated relationship between parts of the bits of code blocks transmitted in sequence during the channel polarization for encoding. In addition, the invention further provides two decoding methods, which comprise the following steps of: decoding by using a simple and rapid successive cancellation (SC) algorithm, and performing iterative decoding by using a Tanner-graph-based belief propagation algorithm with excellent performance. On the basis of the innovative structure encoder, the encoding and decoding methods provided by the invention are stronger in error correction capability under the condition of not increasing the decoding complexity, and the transmission performance is obviously improved. The encoding and decoding methods are particularly applicable to an actual communication engineering system and have a good popularization and application prospect.

The probability of a computer file being malware is inferred by iteratively propagating domain knowledge among computer files, related clients, and/or related source domains. A graph is generated to include machine nodes representing clients, file nodes representing files residing on the clients, and optionally domain nodes representing source domains hosting the files. The graph also includes edges connecting the machine nodes with the related file nodes, and optionally edges connecting the domain nodes with the related file nodes. Priors and edge potentials are set for the nodes and the edges based on related domain knowledge. The domain knowledge is iteratively propagated and aggregated among the connected nodes through exchanging messages among the connected nodes. The iteration process ends when a stopping criterion is met. The classification and associated marginal probability for each file node are calculated based on the priors, the received messages, and the edge potentials associated with the edges through which the messages were received.

The invention discloses a regional depth edge detection and binocular stereo matching-based three-dimensional reconstruction method, which is implemented by the following steps: (1) shooting a calibration plate image with a mark point at two proper angles by using two black and white cameras; (2) keeping the shooting angles constant and shooting two images of a shooting target object at the same time by using the same camera; (3) performing the epipolar line rectification of the two images of the target objects according to the nominal data of the camera; (4) searching the neighbor regions of each pixel of the two rectified images for a closed region depth edge and building a supporting window; (5) in the built window, computing a normalized cross-correlation coefficient of supported pixels and acquiring the matching price of a central pixel; (6) acquiring a parallax by using a confidence transmission optimization method having an acceleration updating system; (7) estimating an accurate parallax by a subpixel; and (8) computing the three-dimensional coordinates of an actual object point according to the matching relationship between the nominal data of the camera and the pixel and consequently reconstructing the three-dimensional point cloud of the object and reducing the three-dimensional information of a target.

In accordance with at least some embodiments of the present disclosure, a processor for performing stereo matching of a first image and a second image is described. The processor may include a first pipeline stage configured to generate data costs associated with a first tile selected from the first image, wherein the data costs is generated based on pixels in the first tile and corresponding pixels in the second image. The processor may include a second pipeline stage configured to generate disparity values associated with the first tile and an outbound message from the first tile to one of neighboring tiles in the first image, wherein the disparity values and the outbound message are generated based on the data costs and inbound messages from the neighboring tiles to the first tile. The processor may further include a third pipeline stage configured to store the disparity values and the outbound message in a memory, wherein the outbound message is used by the second pipeline stage as one of the inbound messages during processing of a second tile selected from the first image.

The invention relates to a color image three-dimensional reconstruction method based on three-dimensional matching, comprising the following steps of: (1) simultaneously and respectively taking an image from proper angles by using two color cameras; (2) respectively calibrating the internal parameter matrixes and the external parameter matrixes of the two cameras; (3) carrying out polar line correction and image transformation according to calibrated data; (4) working out matching cost for each pixel point in the two corrected images by applying a self-adaption weight window algorithm and acquiring an initial parallax image; (5) marking the reliability coefficient of the pixel initial matching result by adopting matching cost reliability detection and left and right consistency verification; (6) carrying out color segmentation on the images through a Mean-Shift algorithm; (7) carrying out global optimization by a selective confidence propagation algorithm on the basis of color segmentation and pixel reliability classification results to obtain a final parallax image; and (8) working out the three-dimensional coordinates of actual object points on the images according to the calibrated data and the matching relation, thereby reconstructing the three-dimensional point cloud of an object.

A statistical formulation estimates two-dimensional human pose from single images. This is based on a Markov network and on inferring pose parameters from cues such as appearance, shape, edge, and color. A data-driven belief propagation Monte Carlo algorithm performs efficient Bayesian inferencing within a rigorous statistical framework. Experimental results demonstrate the effectiveness of the method in estimating human pose from single images.

A decoder of LDPC codewords on GF(r^q), using the iterative belief propagation algorithm comprises at least storing means to store a posteriori information on variables. Updating means updates the a posteriori information on variables, and computation means computes variables to constrain messages from a posteriori information on variables and variable to constraint messages from previous iteration. Computation means computes a constraint to variable messages from variable to constraint messages computed by the first computation means. Computation means updates the a posteriori information on variables. Shuffle means transfers the a posteriori information on variables to the first computation means, and shuffle means transfers information from the second computation means to the storing means. Compression-storage-decompression means constraint to variable messages. In one form, the invention concerns also corresponding method, computer program and system.

A low density parity check (LDPC) code that is particularly well adapted for hardware implementation of a belief propagation decoder circuit is disclosed. The LDPC code is arranged as a macro matrix (H) whose rows and columns represent block columns and block rows of a corresponding parity check matrix (H_pc). Each non-zero entry corresponds to a permutation matrix, such as a cyclically shifted identity matrix, with the shift corresponding to the position of the permutation matrix entry in the macro matrix. The block columns of the macro matrix are grouped, so that only one column in the macro matrix group contributes to the parity check sum in any given row. The decoder circuitry includes a parity check value estimate memory which may be arranged in banks that can be logically connected in various data widths and depths. A parallel adder generates extrinsic estimates that are applied to parity check update circuitry for generating new parity check value estimates. These parity check value estimates are stored back into the memory, and are forwarded to bit update circuits for updating of probability values for the input nodes. Variations including parallelism, time-sequencing of ultrawide parity check rows, and pairing of circuitry to handle ultrawide code rows, are also disclosed.

The computational efficiency of Finite Element Methods (FEM) on parallel architectures is typically severely limited by sparse iterative solvers. Standard iterative solvers are based on sequential steps of global algebraic operations, which limit their parallel efficiency, and prior art techniques exploit sophisticated programming techniques tailored to specific CPU architectures to improve performance. The inventors present a FEM Multigrid Gaussian Belief Propagation (FMGaBP) technique that eliminates global algebraic operations and sparse data-structures based upon reformulating the variational FEM into a probabilistic inference problem based upon graphical models. Further, the inventors present new formulations for FMGaBP, which further enhance its computation and communication complexities where the parallel features of FMGaBP are leveraged to multicore architectures.

In an advanced adaptive modulation and coding (AMC) scheme, the code rate and the parity-check matrix (PCM) for low-density parity-check (LDPC) codes are adapted according to modulation formats and variable-iteration receivers. The degree distribution for the PCM adaptation is designed by heuristic optimization to minimize the required SNR via an extrinsic information transfer (EXIT) trajectory analysis for finite-iteration decoding. The method uses dynamic window decoding by generating spatially coupled PCM for quasi-cyclic LDPC convolutional coding. The method also provides a way to jointly optimize labeling and decoding complexity for high-order and high-dimensional modulations. The problem to use a large number of different LDPC codes for various modulation formats and variable-iteration decoding is also dealt with by linearly dependent PCM adaptation across iteration count to keep using a common generator matrix. This PCM adaptation can improve a convergence speed of belief propagation decoding and mitigate an error floor issue.

The invention relates to low density parity check decoding. A method for decoding an encoded data block is described. Decoding is performed in a pipelined manner using a layered belief propagation technique and scalable resources, which are configurable to accommodate at least two codeword lengths and at least two code rates. A computer program product, apparatus and device are also described.

The invention discloses a binocular vision positioning and three-dimensional mapping method for an indoor mobile robot. A binocular vision positioning and three-dimensional mapping system for the indoor mobile robot is mainly composed of a two-wheeled mobile robot platform, an odometer module, an analog camera, an FPGA core module, an image acquisition module, a wireless communication module, a storage module, a USB module and a remote data processing host computer. The FPGA core module controls the image acquisition module to collect information of a left image and a right image and sends the information of the left image and the right image to the remote data processing host computer. Distance information between images and the robot is obtained on the remote data processing host computer according to the information of the left image and the right image in a fast belief propagation algorithm, a three-dimensional environmental map is established in the remote data processing host computer, recognition of a specific marker is realized in an improved SIFT algorithm, and then the position of the mobile robot is determined through information integration in a particle filter algorithm. The binocular vision positioning and three-dimensional mapping system is compact in structure and capable of having access to an intelligent space system so that the indoor mobile robot can detect the environment in real time and provide more and accurate services according to the corresponding position information at the same time.

A block of symbols are decoded using iterative belief propagation. A set of belief registers store beliefs that a corresponding symbol in the block has a certain value. Check processors determine output check-to-bit messages from input bit-to-check messages by message-update rules. Link processors connect the set of belief registers to the check processors. Each link processor has an associated message register. Messages and beliefs are passed between the set of belief registers and the check processors via the link processors for a predetermined number of iterations while updating the beliefs to decode the block of symbols based on the beliefs at termination.

Advantageous underwater acoustic (UWA) apparatus, systems and methods are provided according to the present disclosure. The apparatus, systems and methods employ nonbinary low density parity check (LDPC) codes that achieve excellent performance and match well with the underlying modulation. The nonbinary LDPC codes of the proposed apparatus, systems and methods are formed, at least in part, from a generator matrix that has a high density to reduce the peak-to-average-power ratio (PAPR) with minimal overhead. The disclosed apparatus, systems and methods employ nonbinary regular LDPC cycle codes if the constellation is large and nonbinary irregular LDPC codes if the constellation is small or moderate. The nonbinary irregular and regular LDPC codes enable: i) parallel processing in linear-time encoding; ii) parallel processing in sequential belief propagation decoding; and iii) considerable resource reduction on the code storage for encoding and decoding.

Systems and methods for performing image editing operations may divide an input image into overlapping patches and assign those patches to locations in a reconstructed output image such that visual artifacts are minimized. The methods may use belief propagation to compute a joint probability for the assignment of active patch labels to output image nodes. The computation may include an exclusivity term, steering the solution such that each patch is preferably only used once in the output image. The methods may include a pre-computation of a pruned list of candidate patches for placing next to each patch in the output image, dependent on local evidence (e.g., color, intensity, or user-driven placement) for each patch. The pre-computation may include determining groupings of patches, each forming a highly compatible loop of neighboring patches for a given candidate patch. The methods may be implemented as program instructions executable by a CPU and/or GPU.

In this invention, a new class of finite precision multilevel decoders for low-density parity-check (LDPC) codes is presented. These decoders are much lower in complexity compared to the standard belief propagation (BP) decoder. Messages utilized by these decoders are quantized to certain levels based on the number of bits allowed for representation in hardware. A message update function specifically defined as part of the invention, is used to determine the outgoing message at the variable node, and the simple min operation along with modulo 2 sum of signs is used at the check node. A general methodology is provided to obtain the multilevel decoders, which is based on reducing failures due to trapping sets and improving the guaranteed error-correction capability of a code. Hence these decoders improve the iterative decoding process on finite length graphs and have the potential to outperform the standard floating-point BP decoder in the error floor region. The description and apparatus of 3-bit decoders for column-weight three LDPC codes is also presented.

The invention discloses a quasi-cyclic low density odd-even check code belief propagation decoding method based on oscillation. In the initial iteration, a mean value is used for obtaining normalizing factors alpha and beta, under different signal-to-noise ratios, an extrinsic information value of a check node is dynamically corrected to reduce fluctuation of the extrinsic information value of the check node, the quasi-cyclic low density odd-even check code belief propagation decoding method unites a bit flip method in hard decision decoding to reduce occurrences of the oscillation, decoding performance of transmission data can be improved, and decoding complexity can be reduced. Decoding methods of soft decision and hard decision are combined to obtain big decoding gains. Balancing states of complexity and decoding efficiency of a decoder are achieved due to the facts that combination of low density odd-even check code hard decision decoding and low density odd-even check code soft decision decoding and excellent coding gains of a soft decision belief propagation algorithm are utilized, and a hard decision decoding method is fully applied to reduce influences of accumulated error code words.

The invention discloses a low-complexity belief propagation detection algorithm for a large-scale MIMO system. A corresponding factor graph is built by utilizing an equivalent real number field model, and complex number field operation is translated into real number field operation, thereby implementing BP-based iteration detection; wherein the factor graph is used for representing a dependency relation between a receiving signal and a transmitting signal, the transmitting signal is utilized as a signal node, and the receiving signal is utilized as an observation node; each signal node updates prior information according to posterior information obtained from the observation node, and then transmits the updated prior information to all observation nodes connected with the signal node; each observation node calculates updates posterior information according to prior information obtained from the signal node, and then transmits the updated posterior information to a signal node connected with the observation node. According to the low-complexity belief propagation detection algorithm for the large-scale MIMO system, a symbol-based large-scale MIMO detection algorithm is implemented; high-dimensional matrix inversion is avoided, and the low-complexity belief propagation detection algorithm can be greatly suitable for application scenarios of the large-scale MIMO.

The invention discloses a LDPC code linear programming decoding method based on a minimum polyhedral model and mainly solves low decoding speed in conventional LDPC code linear programming decoding and an erroneous platform in information transmission decoding. The method comprises steps of: slacking the maximum likelihood decoding of a LDPC code into a linear programming LP model based on a minimum polyhedron by using a verification node decomposition method; establishing an augmented Lagrange function by using the sparsity and the orthogonality of a matrix in the LP model based on the minimum polyhedron, and carrying out iteration by using an alternating direction method of multipliers (ADMM) to solve decoded codes. Compared with a LP decoding method based on the ADMM, the LDPC code linear programming decoding method is increased in decoding speed while not decreasing LP decoding performance, and does not generate an erroneous platform in a high signal-to-noise ratio compared with a belief propagation (BP) decoding method. The LDPC code linear programming decoding method can be used in the technical field of communication to increase the efficiency of a communication system decoding module.