131 results about "Residual coding" patented technology

The invention discloses a training method, a segmentation method and a device of a multi-mode weak supervision medical image segmentation model. The method comprises the following steps: unifying the size of a medical image; in the first stage of the network model, CAM graphs under different scales are extracted through a residual coding network, and single-mode CAM graphs under different scales are merged through a Sigmoid activation function; fusing CAMs of different modes; encoding a foreground image, a background image, a text label corresponding to the foreground image and a text label corresponding to the background image by using CLIP, calculating the similarity between an image encoding vector and a corresponding text encoding vector, and maximizing the similarity as a loss item; meanwhile, taking an opposite number of an absolute value of a difference value between the CAM of the foreground image and the CAM of the background image as one of loss items to train a model; performing foreground and background segmentation on the final CAM by taking 0.5 as a threshold value, and calculating a final Dice score with a real label so as to evaluate the performance of the model; according to the method, the multi-modal CAM fusion problem can be solved by utilizing the strong representation capability of the large model.

To provide a method and apparatus for improving image coding efficiency.SOLUTION: An image decoding method includes the steps of: acquiring a sign data hiding enable flag; acquiring a transform skip residual coding (TSRC) enable flag based on the sign data hiding enable flag; acquiring residual information for a current block based on the TSRC enable flag; deriving a residual sample of the current block based on the residual information; and generating a reconstructed picture based on the residual sample. The sign data hiding enable flag is a flag indicating whether sign data hiding is enabled, the TSRC enable flag is a flag indicating whether TSRC is enabled, and the TSRC enable flag is acquired based on the sign data hiding enable flag having a value of zero.SELECTED DRAWING: Figure 10

A method and device for decoding an encoded data propose to determine a residual error image based on the decoded image only and generated by an invertible neural network without needed any additional signaling data. The residual error image is combined with the decoded image, resulting into an enhanced image. The invertible neural network is trained to determine, in forward pass, a latent variable from a decoded image and a residual coding map and to determine, in backward pass, a reconstructed residual map from a decoded image and a random residual map, so that the latent variable follows a normal distribution or a uniform distribution.

The invention discloses an electric energy meter data compression method and system, electronic equipment and a storage medium, and the method comprises the steps: carrying out the first-order differential processing of an electric quantity measurement data sequence, obtaining a residual data sequence, greatly reducing the value domain of the residual data sequence, greatly improving the repetition, and remarkably reducing the statistical entropy; low-entropy and high-redundancy ideal input is created for subsequent coding, and the compression ratio is integrally improved. According to the method, the residual data sequence is subjected to coding processing, positive and negative information is reserved, the problem that complement storage equivalent entropy is not reduced is avoided, bit column recombination processing or byte column recombination processing is performed on the residual coding sequence, and a large number of continuous data streams of local repeated fragments are generated, so that in an electric energy meter environment in which the MCU computing power and the RAM are limited subsequently, the residual data sequence is subjected to bit column recombination processing or byte column recombination processing. The lightweight compression algorithm can also quickly discover and reuse redundancy in an extremely short window, and the overall compression efficiency is remarkably improved.

A method of decoding an image according to a decoding apparatus of the present document includes the steps of obtaining a dependent quantization available flag regarding whether dependent quantization is available, obtaining a sign data hiding available flag regarding whether sign data hiding is available, and obtaining a transform skip residual coding (TSRC) available flag regarding whether TSRC is available based on the dependent quantization available flag and the sign data hiding available flag.

Methods, apparatuses, and non-transitory computer-readable storage mediums are provided for video encoding and decoding. An encoder may signal a control flag in a slice header level. The encoder may signal at least one syntax element in the slice header level. The control flag may be a residual coding disabled flag. The at least one syntax element may comprise a syntax element related to a residual coding rice index.

The present disclosure provides systems and methods for signaling a maximum transform size and a residual coding method. According to certain disclosed embodiments, the methods include: receiving a bitstream comprising a set of pictures, determining, according to the received bitstream, a value of a coded tree block size; and determining, based on the value of the coded tree block size, whether to signal a flag indicating a maximum transform size for luma samples.

This invention provides a system and method for encoding and decoding video data. [Solution] The solution includes receiving a video frame for residual coding, determining whether the video frame is coded according to a first coding mode, and turning off sign data hiding for residual coding in response to determining whether the video frame is coded according to the first coding mode.

The application discloses a point cloud data processing method, system, device and medium of a humanoid robot. The method comprises the following steps: acquiring a motion state parameter of a robot and initial point cloud data collected; performing dynamic sampling processing on the robot according to the motion state parameter and the initial point cloud data, and generating a projection image; performing multi-channel feature extraction and dynamic weight fusion processing on the projection image, and obtaining an encoded feature map; performing multi-scale residual coding processing on the encoded feature map according to an information theory constraint, and obtaining encoded data; and performing reconstruction processing on the encoded data according to the motion state parameter, and obtaining reconstructed point cloud data. The embodiment of the application can improve the processing efficiency of point cloud data and can be widely applied to the technical field of intelligent robots.

A residual coding method is provided. The method comprises: when the current frame is an inter-frame prediction frame, determining whether a residual of a background region in the current frame needs to be compensated for; and when the residual of the background region in the current frame needs to be compensated for, coding first residual data, and when the residual of the background region in the current frame does not need to be compensated for, coding second residual data, wherein the first residual data is residual data of an entire image of the current frame, and the second residual data is residual data of a target region in the current frame. The present disclosure further provides a video coding method using the residual coding method, and corresponding devices and a storage medium.

The invention discloses a breast lesion image semi-supervised segmentation method. The method belongs to the technical field of medical image processing and semi-supervised learning. The technical problem of improving the feature extraction capability of a semi-supervised segmentation network in a complex breast lesion segmentation scene under a small number of labeled samples and realizing accurate segmentation of a breast lesion is solved. A dual-scale image of a mammary gland image is generated, so that the network can learn features of mammary gland lesions from different scales, the ability of the network to extract features of mammary gland lesions with large size and form differences is improved, data perturbation is performed on the dual-scale image, and a generator performs mutual generation on segmentation confidence maps of a source image and a perturbation image. And the network stability is improved. Through cascade inverse residual coding feature fusion and decoding feature fusion, more potential features are mined by using existing features, the segmentation performance of the network is improved under a small number of labeled samples, and accurate segmentation of a breast lesion can be realized by inputting a breast image into a final segmentation network.

An image encoding / decoding method and apparatus are provided. An image decoding method according to the present disclosure may comprise obtaining a reconstructed picture, determining a target boundary of deblocking filtering in the reconstructed picture, determining a boundary strength for the target boundary, and applying deblocking filtering to the target boundary based on the boundary strength. Based on the target boundary being a transform block boundary and a color component of the reconstructed picture being a chroma component, the boundary strength may be determined based on whether joint CbCr residual coding is performed on at least one of two blocks adjacent to the target boundary, and the joint CbCr residual coding may correspond to encoding residual samples for a chroma Cb component and a chroma Cr component as a single transform block.

The invention relates to an adaptive configuration method for compression parameters of ordered key learning. The adaptive configuration method comprises the following steps: selecting a public data set containing ordered integer keys; calculating a gap value between two adjacent integer keys so as to obtain a gap value sequence set between all the integer keys; all change point positions of the sequence set are calculated by adopting a Bayesian online change point detection algorithm and mapped back to the original data set, and the original data set is divided into a plurality of partitions with similar statistical characteristics by change points; setting an initial error upper limit threshold value of each partition, and constructing a piecewise linear function of each partition and a corresponding bit-level space total cost function; solving an optimal error upper limit and a corresponding residual bit width in combination with the partition variance and the partition constraint; constructing a configuration portrait library containing optimal configuration; and traversing all partitions, generating a compression parameter adaptive configuration table, constructing a learning compressor according to the compression parameter adaptive configuration table, performing error controlled prediction and residual encoding on an original sequence, and outputting a final compression result.

Image decoding method, image encoding method, and transmission method. An image decoding method implemented by a decoding apparatus according to the present document includes the steps of: obtaining prediction-related information of a current block; obtaining prediction-related information of a current block; obtaining a dependent quantization enabled flag related to whether dependent quantization is available; obtaining a transform skip residual coding (TSRC) enabled flag related to whether TSRC is available based on the dependent quantization enabled flag; obtaining residual information of a residual coding syntax of the current block derived based on the TSRC enabled flag; deriving motion information of the current block based on the prediction-related information; deriving prediction samples of the current block based on the motion information; deriving residual samples of the current block based on the residual information; and generating a reconstructed picture based on the prediction samples and the residual samples.

The invention discloses an air conditioner fault early warning system and method based on deep learning, and the method comprises the following steps: S1, collecting data, forming current multi-channel time sequence input data, and obtaining a historical fault-free operation data sample; s2, constructing a normal state prototype sample; s3, generating prototype residual error coding features; s4, inputting the current multi-channel time sequence input data and the prototype residual encoding features into the improved multi-scale convolution branch network to obtain original feature branches and residual feature branches, and introducing a causal-guided convolution kernel selection mechanism; s5, fusing the original feature branch and the residual feature branch; s6, obtaining an abnormal score value based on fusion feature representation; s7, the abnormal score value is compared with a preset threshold value, and an air conditioner fault early warning instruction is output; and S8, when the abnormal score value does not reach the preset threshold value, updating the prototype sample in the normal state. The air conditioner fault early warning method improves the accuracy and self-adaptive capacity of air conditioner fault early warning, and is suitable for intelligent operation and maintenance scenes of an air conditioner system.

SUMMARY The present invention provides systems and methods for signaling a maximum transform size and residual coding methods. According to certain disclosed embodiments, the method comprises: receiving a bitstream comprising a set of pictures; determining, from the received bitstream, a value of a coding tree block size; and based on the value of the coding tree block size, determining whether to signal a flag indicating a maximum transform size of luma samples.

The invention provides an image decoding apparatus, an encoding apparatus, and an image information transmitting apparatus. A method for decoding an image by a decoding device according to the present document comprises the steps of: acquiring a dependent quantization availability flag regarding whether dependent quantization is available; obtaining a symbol data hiding available mark about whether symbol data hiding is available or not; and obtaining a transform skip residual coding (TSRC) available flag on whether TSRC is available based on the quantization dependent available flag and the symbol data hidden available flag.

Methods and apparatuses are provided for video coding with lossless coding modes. The method includes: determining whether a current block is in a transform skip mode; in response to determining that the current block is in the transform skip mode, determining whether a first residual coding scheme or a second residual coding scheme is selected for the current block in the transform skip mode; and signaling a first indication which indicates whether a current block is in the transform skip mode and signaling a second indication which indicates whether the first residual coding scheme or the second residual coding scheme is applied for blocks in the transform skip mode.

In a method of video decoding, a no transform skip residual coding rice constraint flag in a general constraint information (GCI) syntax structure for coded video data in a bitstream is determined. The no transform skip residual coding rice constraint flag is associated with a residual coding tool. When the no transform skip residual coding rice constraint flag in the GCI syntax structure is equal to 1, the coded video data is decoded without invoking the residual coding tool. When the no transform skip residual coding rice constraint flag is equal to 0, a value of a transform skip residual coding rice present in slice header flag for one of one or more coded layer video sequences (CLVSs) in the coded video data is determined. The coded video data in the bitstream is determined based on the value of the transform skip residual coding rice present in slice header flag.

Methods and apparatuses for video decoding and video encoding and a non-transitory computer-readable storage medium are described. The apparatus for video decoding includes processing circuitry configured to receive a bitstream that includes a syntax element. The syntax element indicates whether at least one residual related syntax element is signaled in the bitstream for a current block in a current picture when the current block is coded according to a prediction block in the current picture and an offset between the current block and the prediction block is indicated by a block vector (BV). When the syntax element indicates that no residual related syntax element is signaled, the processing circuitry is configured to reconstruct the current block by directly copying values of reconstructed samples in the prediction block.

The application relates to the technical field of data compression storage, and discloses a time series database driven monitoring data compression storage method, which comprises the following steps: S1, preprocessing multi-source monitoring data, and constructing a dynamic tensor containing a timestamp, a numerical index and a multi-dimensional label; S2, performing dynamic dimension reduction processing on the dynamic tensor to generate a core tensor and a multi-dimensional factor matrix; S3, based on the core tensor and the multi-dimensional factor matrix, determining compression parameters, decompression parallelism and index granularity through a joint optimization model; and S4, performing hierarchical coding on residual data generated by the dynamic dimension reduction processing to generate a light residual coding result. Through dynamic tensor decomposition and incremental updating technology, low storage overhead and real-time dimension expansion capability of streaming monitoring data are realized, the problems of calculation redundancy and storage expansion caused by the incapability of the streaming monitoring data to adapt to dynamic newly-added labels are solved, and the frequent reconstruction cost caused by dynamic expansion of the data is avoided.

An example method of encoding a point cloud includes determining that residual values of all but one component of an attribute of a point in the point cloud are equal to zero, determining, based on the determination that the residual values of all but the one component of the attribute are equal to zero, that a value of the one component is equal to a magnitude of a residual value of the one component of the attribute minus an offset, encoding the value of the one component, and signaling the encoded value in a bitstream.

Methods, devices, and non-transient, computer-readable storage media are provided for video decoding. In one method, a decoder receives a rice extension flag from the Sequence Parameter Set (SPS) indicating whether an extension of the rice parameter derivation for abs_remainder and dec_abs_level binarization is enabled. In a second method, the decoder can receive an enabled rice adaptation flag from the Sequence Parameter Set (SPS) indicating whether the rice parameter derivation for abs_remainder and dec_abs_level binarization is initialized at the beginning of each transform unit (TU) with accumulated statistics from previous TUs.

For a transform block of a frame of a video encoded with conventional residual coding, if an inferred sub-block of a transform block of an sb codeflag is not present is a DC sub-block and / or the last one of transform blocks containing a non-zero coefficient level, the decoder inferes that the sb codeflag of that sub-block is 1. And otherwise, inferring that the sb coded flag is 0. If a transform block is encoded using transform skip residual encoding, the sb coded flag of a sub-block in which no sb coded flag exists is estimated to be 1. The decoder determines an initial context value of an entropy coding model based on the sub-block flag of the inferred sub-block, and determines a sub-block flag of the encoded sub-block using the entropy coding model.

An image decoding method performed by a decoding apparatus according to the present document comprises the steps of: obtaining a sign data hiding availability flag indicating whether or not sign data hiding is available for a current slice; obtaining a TSRC availability flag indicating whether or not TSRC is available for a transform skip block of the current slice; obtaining residual coding information about the transform skip block on the basis of the TSRC availability flag; deriving a residual sample for the transform skip block on the basis of the residual coding information; and generating a reconstruction picture on the basis of the residual sample, wherein the TSRC availability flag is obtained on the basis of the sign data hiding availability flag.

The invention relates to the field of urban road traffic pollution assessment, in particular to a city heavy diesel emission pollution field rapid description method based on dynamic point distribution, comprising the following steps: S1, constructing joint input data of road emission spatial features and meteorological conditions; s2, extracting multi-scale features of an emission space based on a residual coding structure; s3, performing adaptive modulation on the multi-scale spatial features according to meteorological conditions; s4, establishing a mapping relationship between the conditionally modulated high-dimensional spatial features and the optimal receptor stationing strategy by using a deep learning model; and the like. Compared with the prior art, on the premise that the spatial resolution of the pollution concentration field is guaranteed, the set number of pollution concentration acceptor points is remarkably reduced, and the overall efficiency of road traffic pollution emission simulation and pollution hot spot recognition is improved; an efficient and reliable technical means is provided for urban traffic environment impact assessment, pollution prevention and control decision and refined traffic management.

SUMMARY The present invention provides systems and methods for signaling a maximum transform size and residual coding methods. According to certain disclosed embodiments, the method comprises: receiving a bitstream comprising a set of pictures; determining, from the received bitstream, a value of a coding tree block size; and based on the value of the coding tree block size, determining whether to signal a flag indicating a maximum transform size of luma samples.

This invention discloses a transformer-based video decoding and filtering method, system, terminal, and storage medium. The method includes: acquiring a reference frame and a current frame; passing the reference frame and the current frame through a motion estimation network to obtain motion vectors; performing motion encoding and decoding on the motion vectors to obtain decoded motion vectors; performing motion compensation on the decoded motion vectors to obtain multi-scale context information; inputting the current frame and the multi-scale context information into a residual coding network to obtain a feature representation of the current frame; inputting the feature representation and reference features of the reference frame into a preprocessing module of a temporal context filter for preprocessing to obtain key vectors, value vectors, and query vectors constituting an attention mechanism; inputting the key vectors, value vectors, and query vectors into the filtering main module of the temporal context filter for filtering optimization; outputting a final reconstructed frame; and participating the final reconstructed frame in subsequent video decoding and filtering processes. This invention significantly improves the quality of reconstructed video.

The application discloses a breast lesion image semi-supervised segmentation method, and belongs to the technical field of medical image processing and semi-supervised learning. The technical problem of improving the feature extraction capability of a semi-supervised segmentation network in a complex breast lesion segmentation scene and realizing accurate segmentation of a breast lesion is solved. A double-scale image of a breast image is generated, so that the network can learn the features of the breast lesion from different scales, and the feature extraction capability of the network for the breast lesion with a large size and shape difference is improved. Meanwhile, data disturbance is performed on the double-scale image, the segmentation confidence maps of the source image and the disturbed image are generated by the generator, and the stability of the network is improved. Through cascading inverse residual coding feature fusion and decoding feature fusion, more potential features are mined from existing features, the segmentation performance of the network is improved under a small amount of labeled samples, and accurate segmentation of the breast lesion can be realized by inputting the breast image into the final segmentation network.