41 results about "Multiple hypothesis" patented technology

The invention relates to a robot autonomous operation optimization method and system based on virtual simulation and multi-hypothesis planning. The method comprises the steps that a virtual simulation platform scene with the same configuration as a real experiment scene is built; robot training is carried out in the virtual simulation platform scene, and an autonomous operation strategy is generated; constructing a plurality of hypothetical states based on the current state of the robot, and obtaining a virtual track action sequence of each hypothetical state based on an autonomous operation strategy; sequentially verifying whether the virtual track action sequence corresponding to each virtual state is feasible or not according to a time sequence, and if the virtual track action sequence is feasible, taking the virtual track action sequence as an optimal virtual track action sequence; and aligning the real robot state with the assumed state corresponding to the optimal virtual track action sequence, and generating a real robot action sequence based on the current state and the optimal virtual track action sequence by adopting a cooperative action integration mechanism. Compared with the prior art, the method has the advantages that the difference between virtuality and reality of the model is reduced, and the execution success rate is increased.

An example device for decoding video data includes one or more processors configured to: determine that a first weight and a second weight are specified for a bi-prediction mode predicted current block of video data; determine whether the current block is to be predicted using multi-hypothesis prediction (MHP) mode with the bi-prediction mode as a base mode; in response to determining that the current block is to be predicted using the MHP mode with the bi-prediction mode as the base mode, determine an additional inter-prediction mode of the MHP mode; generate a first prediction block according to the bi-prediction mode; generate a second prediction block according to the additional inter-prediction mode; generate a final prediction block for the current block according to the MHP mode using the first prediction block and the second prediction block; and decode the current block using the final prediction block.

The invention discloses a space debris target detection and positioning method based on dynamic programming and multi-hypothesis testing, and the method comprises the following steps: carrying out the preprocessing of an obtained space debris image, and obtaining a candidate point state set under different time frames; in the preprocessing, a denoising segmentation extraction channel and a lightweight CNN target extraction channel are divided according to a statistical local signal-to-noise ratio; obtaining a plurality of motion paths corresponding to each candidate point according to the candidate point state sets under different time frames; and for each candidate point, generating an initial hypothesis set according to the corresponding multiple motion paths, and processing the initial hypothesis set through Kalman filtering to obtain the position of the candidate point. According to the method, the detection precision of the dark and weak target is remarkably improved, the false alarm rate is reduced, the real-time performance of detection is improved, and the method is suitable for real-time monitoring scenes. The research result provides effective technical support for improving the operation safety of the spacecraft.

The invention discloses an active sonar target identification method based on message passing multi-hypothesis tracking, and belongs to the technical field of underwater target detection. The method comprises the following steps: firstly, processing single-cycle sonar data by using adaptive constant false alarm detection, obtaining observation points, and forming an effective observation set through hierarchical clustering; thirdly, extrapolating the existing target, calculating the association probability of the existing target and the current observation by adopting a message passing method, and generating a plurality of groups of association hypotheses; then, the target state is corrected through Kalman filtering, target rejection and new management are executed, and a high-probability hypothesis is reserved; and finally, estimating the motion speed based on the target multi-period position sequence, and distinguishing a static object from a suspicious motion target according to a speed threshold. Through the multi-hypothesis tracking and probability message passing mechanism, clutter interference and static target influence in a complex underwater environment are effectively overcome, and the stability of target tracking, the accuracy of speed estimation and the overall reliability of system identification are remarkably improved.

A multi-hypothesis hybrid video coding and decoding method and apparatus for cross component model merge mode. According to the method, input data associated with a current block is received, the current block comprises a first color block and a second color block, and the input data comprises pixel data to be encoded at an encoder end or encoded data to be decoded at a decoder end and associated with the current block. A base predictor for a second color block is determined, where the base predictor corresponds to an intra predictor or an inter predictor. One or more hypotheses of a second color block are determined, wherein each of the one or more hypotheses corresponds to one cross-component prediction candidate. A final predictor is derived by mixing the base predictor and the one or more hypotheses. And encoding or decoding the second color block using prediction data including the final predictor.

The invention discloses a signal sorting method and device based on multi-hypothesis tracking, and a storage medium, and relates to the technical field of radar signal processing. The method comprises the following steps: acquiring a pulse TOA sequence, and calculating a PRI candidate value set; searching an effective initial pulse pair in the TOA sequence, and creating an initial tracking hypothesis for the effective initial pulse pair; aiming at the initial tracking hypothesis, searching the next matched pulse; if the matched pulse is found, judging whether the currently used PRI value is consistent with the PRI value of the initial tracking hypothesis or not: if so, updating the initial tracking hypothesis; if not, creating a new branch hypothesis; fusing and pruning all tracking hypotheses at regular intervals; confirming the tracking hypothesis meeting the output condition as a sorted radar pulse sequence, and outputting the sorted pulse sequence and a corresponding PRI change mode sequence; and repeating the steps in sequence for the residual pulse sequence which is not sorted until the sorting ending condition is met. According to the invention, the reliability of PRI sorting is improved.

Methods and apparatus of video coding using new Sample-based Prediction Offset (SPO) are disclosed. According to one method, SPO for coding systems with multiple hypothesis is derived based on a template area of the current block and PRW (Position-Related Weighting) to refine prediction. In one embodiment, SPO can be applied to the final prediction. In another embodiment, SPO can be applied to each of the multiple hypothesis or some of them. In yet another method, the SPO is applied to the final prediction, or each of the multiple hypothesis or some of them selectively. In yet another method, the SPO and OBMC (Overlapped Block Motion Compensation) are both applied to a coding block.

The method performed by at least one processor of the video decoder includes receiving a coded video bitstream including at least one picture encoded according to multiple hypothesis arithmetic coding and one or more syntax elements. The method further includes decoding each syntax element from the one or more syntax elements based on the multiple hypothesis arithmetic coding. The method further includes selecting a probability update rate from a plurality of probability update rates based on a predetermined condition, the plurality of probability update rates including a first probability update rate higher than a second probability update rate. The method further includes updating at least one probability model utilized in the multiple hypothesis arithmetic coding based on the selected probability update rate. The method further includes decoding at least one block in the at least one picture based on the decoded one or more syntax elements.

A method and apparatus for hybrid IBC mode. According to the method, a first predictor or a first hypothesis is derived by using a first prediction mode including an Intra Block Copy (IBC) prediction mode or an Intra Template Matching Prediction (IntraTMP) mode. A second predictor is derived. A hybrid predictor is derived by mixing the first predictor or the first hypothesis with the second predictor. And encoding or decoding the current block by using a predictor candidate set including the hybrid predictor. According to another method, for IBC-CIIP (Joint Intra Block Copy and Intra Prediction), behavior between Inter Prediction mode, IBC AMVP mode, and IBC Merge mode remains consistent.

Methods, apparatuses, and non-transitory computer-readable storage mediums are provided for video coding. In one method, a decoder obtains, for a current chroma block, a non-linear model (non-LM) mode and a linear model (LM) mode, and the LM mode includes a cross component linear model (CCLM) mode and a multi-model linear model (MMLM) mode. The decoder then combines the non-LM mode and the LM mode for a multi-hypothesis- based chroma prediction (MCP) for the current chroma block.

The invention provides a multi-hypothesis combining circuit. The multi-hypothesis combining circuit comprises a coherent combining circuit and a selecting circuit. The coherent combination circuit performs coherent combination according to a plurality of correlation operation outputs and a plurality of sign sequences so as to generate a plurality of coherent combination outputs, and the plurality of correlation operation outputs respectively correspond to a plurality of channels. And each coherent combination output in the plurality of coherent combination outputs is obtained from the plurality of correlation operation outputs and a sign sequence in the plurality of sign sequences. The selection circuit generates and outputs a combination output of the multi-hypothesis combination circuit according to the plurality of coherent combination outputs.

A method of encoding or decoding video data includes: for a multi-hypothesis prediction (MHP) process, determining a plurality of prediction templates based on a plurality of weights; comparing the plurality of prediction templates with a current template of the current block; determining a weight from the plurality of weights based on the comparison of the plurality of prediction templates and the current template of the current block; determining one or more prediction hypotheses; determining a prediction signal for the current block based on the one or more prediction hypotheses and the weight; and encoding or decoding the current block based on the prediction signal.

Apparatus and methods for multi-hypothesis measurement configuration in RF sensing are described. An apparatus is configured to receive a sensing configuration from a sensing entity, wherein the sensing configuration indicates a total number of sensing hypotheses and at least a first sensing hypothesis and a second sensing hypothesis different from the first sensing hypothesis; the apparatus is further configured to measure a first set of sensing measurement data associated with the first sensing hypothesis and a second set of sensing measurement data associated with the second sensing hypothesis. The apparatus is further configured to transmit, for the sensing entity, at least one of (1) the first set of sensing measurement data and the second set of sensing measurement data, or (2) an indication of the first sensing hypothesis and the second sensing hypothesis.

An example method for decoding video data includes receiving a video bitstream that includes a current block and a syntax element indicating whether each chroma sample of the current block is determined based on one or more luma samples. The method also includes generating a plurality of hypothesis values for predicting a first chroma sample by combining a plurality of neighboring luma samples of a first luma sample that is co-located with the first chroma sample using a plurality of coefficients. The plurality of hypothesis values include a first plurality of hypothesis values calculated based on top and bottom neighboring samples and respective sets of coefficients, and a second plurality of hypothesis values calculated based on left and right neighboring samples and respective sets of coefficients. The method further includes predicting the first chroma sample by combining the first luma sample and the plurality of hypothesis values.

Systems and methods are described for video coding using generalized bi-prediction. In an exemplary embodiment, to code a current block of a video in a bitstream, a first reference block is selected from a first reference picture and a second reference block is selected from a second reference picture. Each reference block is associated with a weight, where the weight may be an arbitrary weight ranging, e.g., between 0 and 1. The current block is predicted using a weighted sum of the reference blocks. The weights may be selected from among a plurality of candidate weights. Candidate weights may be signaled in the bitstream or may be derived implicitly based on a template. Candidate weights may be pruned to avoid out-of-range or substantially duplicate candidate weights. Generalized bi-prediction may additionally be used in frame rate up conversion.

The invention discloses an edge thundersight fusion method and system based on flattening multi-hypothesis tracking, and the method comprises the steps: obtaining the observation data and track list of a plurality of sensors at the current moment, the list comprises a plurality of track objects, and each track object comprises state information, a traceability identification sequence and a confidence score; predicting the state information of each track object based on a preset equation to obtain predicted state information; calculating the correlation degree between the prediction state information of each track object and the observation data, performing hypothesis branch operation based on the correlation degree, and generating at least one cloned track object for the track object with the correlation ambiguity; conflict detection is carried out based on the traceability identification sequence of each track object; determining an optimal hypothesis set based on the detection result and the confidence score of each track object; and fusing the state information belonging to the same target in the set based on a preset weight to obtain target data of each target. According to the invention, the accuracy of the output target data can be improved.

This invention discloses an edge radar-visual fusion method and system based on flattened multi-hypothesis tracking. The method includes: acquiring observation data and a track list from multiple sensors at the current moment, the list including multiple track objects, each track object including state information, a source identification sequence, and a confidence score; predicting the state information of each track object based on a preset equation to obtain predicted state information; calculating the correlation between the predicted state information of each track object and the observation data, and performing hypothesis branching operations based on the correlation to generate at least one clone track object for track objects with ambiguous correlation; performing conflict detection based on the source identification sequence of each track object; determining the optimal hypothesis set based on the detection results and the confidence score of each track object; and fusing the state information of the same target in the set based on preset weights to obtain target data for each target. This invention can improve the accuracy of the output target data.

The invention discloses a prediction method and system for generating a multi-hypothesis intention with confidence and a reasoning process, and belongs to the technical field of artificial intelligence and intelligent health monitoring. The method aims at solving the problems that an existing non-language user intention prediction technology is single in prediction result, opaque in process and lack of credibility evaluation. The method comprises the following steps: acquiring a first visual angle visual data stream and a physiological data stream of a user in real time; converting the visual data into a time sequence semantic observation sequence through a first preset model; analyzing physiological data and extracting state characteristic indexes; fusing the information and the environment context into multi-modal comprehensive situation data; and finally, inputting the context data into a large language model, driving the large language model to generate an ordered multi-hypothesis intention prediction list through a specific prompt engineering strategy, and enabling each hypothesis to comprise a quantized confidence score and a detailed reasoning process text. By generating multiple hypotheses and providing a transparent reasoning process and quantitative confidence, the accuracy, reliability and interpretability of intention prediction are remarkably improved, and more intelligent decision support is provided for nursing personnel.

Methods and apparatuses for processing a current block within a current picture. The processing may include determining whether a second syntax element S1 is present in a coded video bitstream. The processing may include, if the S1 is not present, inferring which one of implicit signaling and explicit signaling is used to signal a second piece of motion information M1, or, if the S1 is present, determining which one of implicit signaling and explicit signaling is used to signal the M1 based on a decoded value of the S1. The processing may include, if implicit signaling is inferred or determined, deriving the M1, or, if explicit signaling is inferred or determined, determining the M1. The processing may include generating a prediction block Pb01 for the current block based on a first piece of motion information M0, the M1, and a weight factor W1. The second syntax element S1 may be a multi-hypothesis prediction, MHP, merge flag and the third syntax element S2 may be a motion information index.

An update parameter modifier for a probability model parameter that controls an update function for a probability model for entropy coding the sequence of syntax elements is determined. The probability model parameter is modified using the update parameter modifier to obtain a modified probability model parameter. A symbol for a syntax element of the sequence is entropy coded using the probability model. The probability model is then updated using the update function with the modified probability model parameter.

A multi-hypothesis positioning method includes: receiving, at an apparatus from a digital twin function, multi-hypothesis positioning assistance data; and determining to implement multi-hypothesis positioning based on at least a subset of the multi-hypothesis positioning assistance data.

A method and apparatus for predicting coding are disclosed. According to the method, combined prediction members are determined, wherein each combined prediction member includes a weighted sum of a first prediction candidate and a second prediction candidate using a target weight selected from a weight set. Boundary matching costs associated with the combined prediction members are determined, wherein each boundary matching cost is determined by using prediction samples of a current block based on the combined prediction member with the target weight and neighboring reconstructed samples of the current block for the combined prediction member with the target weight. A final combined prediction based on at least one boundary matching cost decision is then used to encode or decode the current block.

A method and apparatus for video coding are disclosed. According to the method, a current block is received at an encoder side or compressed data including the current block is received at a decoder side, wherein the current block includes one luma block and one or more chroma blocks, and the current block is generated by partitioning an image region into one or more partition blocks containing the current block using a single partition tree. A target coding mode of the current block is determined. Then, the current block is encoded or decoded according to the target coding mode, wherein if the target coding mode corresponds to a multi-hypothesis prediction mode and a width, a height, or an area of the current block is less than a threshold, additional hypotheses for prediction of the one or more chroma blocks are disabled.

A method of encoding or decoding video data includes determining, for a multiple hypothesis prediction (MHP) process, a plurality of prediction templates based on a plurality of weights; comparing the plurality of prediction templates to a current template of a current block; determining a weight from the plurality of weights based on the comparing of the plurality of prediction templates to the current template of the current block; determining one or more prediction hypotheses; determining a prediction signal for the current block based on the one or more prediction hypotheses and the weight; and encoding or decoding the current block based on the prediction signal.

A method and apparatus for video coding using multi-hypothesis blending for cross-component model merge mode. According to the method, input data associated with a current block comprising a first-colour block and a second-colour block are received, where the input data comprise pixel data to be encoded at an encoder side or coded data associated with the current block to be decoded at a decoder side. A base predictor for the second-colour block is determined, where the base predictor corresponds to an intra predictor or an inter predictor. One or more hypotheses for the second-colour block are determined, where each of said one or more hypotheses corresponds to a cross-component prediction candidate. A final predictor is derived by blending the base predictor and said one or more hypotheses. The second-colour block is encoded or decoded by using prediction data comprising the final predictor.

This invention discloses a method for reverse tracing of submerged oil pollution sources based on knowledge graphs and spatiotemporal big data, relating to the field of environmental monitoring technology. The method includes: collecting multi-source heterogeneous spatiotemporal data and aligning and fusing it to form a fused spatiotemporal dataset; constructing a dynamic spatiotemporal association knowledge graph centered on pollution events based on the fused spatiotemporal dataset to form an initial source-tracing association knowledge subgraph; generating a spatiotemporal backtracking probability field based on a reverse spatiotemporal probability field simulation using multiple hypothesis sets; and inputting the initial source-tracing association knowledge subgraph into a hidden behavior discrimination model trained on a generative adversarial network to identify high-confidence hidden behavior patterns. This invention uses a counterfactual causal inference framework to evaluate and rank the causal effects of associated entities in the converged source-tracing association knowledge graph, obtaining reverse tracing conclusions, thereby achieving accurate, reliable, and interpretable intelligent tracing of submerged oil pollution sources.

A method for predictive coding operates by receiving input data associated with a current block, wherein the input data comprise pixel data for the current block to be encoded at an encoder side or encoded data associated with the current block to be decoded at a decoder side. The method then determines prediction members associated with an existing prediction, wherein each prediction member corresponds to one weighted sum of the existing prediction and at least one target prediction candidate from a group of prediction candidates, and wherein a weight and the at least one target prediction candidate are jointly decided by a joint index. Encoding or decoding of the current block is performed using a final prediction decided based on the joint index.

This document describes a multi-hypothesis-based data fusion tracker. Each hypothesis corresponds to a different pseudo-measurement type. The fusion tracker automatically determines which pseudo-measurement type is more likely to be accurate in the current situation using a predefined error covariance associated with radar. The fusion tracker can rely on either a combination of radar and vision computations, or it can ignore vision-based pseudo-measurements and rely solely on radar pseudo-measurements. By selecting between three different bounding boxes (view-angle-based, view-lateral-position-based, or radar-based only), the fusion tracker can balance accuracy and speed when drawing, repositioning, or resizing bounding boxes, even in congested traffic or other high-volume conditions.

The various implementations described herein include methods and systems for coding video. In one aspect, a video bitstream includes a current coding block of an image frame and signals a first syntax element for a multi-hypothesis cross-component prediction (MH-CCP) mode. A computing system determines that the MH-CCP mode is enabled to reconstruct each of a plurality of chroma samples of the current coding block using a respective luma sample and one or more neighboring luma samples. The computing device determines a number (N) of model parameters used in the MH-CCP mode for at least the current coding block, and identifies one or more neighboring luma samples of a first luma sample based on the number (N) of model parameters. A first chroma sample that is collocated with the first luma sample is determined based on the first luma sample and the one or more neighboring luma samples.