42results about How to "Reduce storage complexity" patented technology

Using macro block type and quantized parameters, the disclosed method determines whether de-block filtering is carried out or not, and determine what kind of macro block filtering mode is selected or adopted in case of de-block filtering is needed. The method does not use residual information of coded macro or information of motion vector to determine whether de-block filtering is needed or not. Based on macro block filtering mode adopted and content of image, de-block filtering procedure is determined in self-adaption. Under precondition of possessing considerable objective and subjective decoded image quality, the disclosed method reduces computation complexity and storage complexity greatly. The invention can be extended to application of other signal processes, not limited in area of video frequency or encoding and decoding images.

A publicly-verifiable cloud data possession checking method is provided. The method includes the three stages of initialization, user challenge-response verification and TTP arbitration; the three stages comprise the following ten steps that: step 1, data segmentation [F->F'] is performed; step 2, verification tag generation [(SK, F)-> HVT] is performed; step 3, data and tag information upload [(F, HVT)->S] is performed; step 4, secret parameters are recorded to a TTP (trusted third party); step 5, the TTP verifies recorded parameters; step 6, [chal] is initiated by a user, and the chal and the verification information of the chal are handed over to the TTP so as to be record; step 7, a server makes a response; step 8, [(R, sk)->(' success', 'failure')] is verified; step 9, relevant information acquisition is verified; and step 10, the trusted third party performs verification publicly. With the publicly-verifiable cloud data possession checking method of the invention adopted, the user can verify the existence and integrity of data in a cloud storage server. The publicly-verifiable cloud data possessions checking method is advantageous in no limitation on the number of user authentication times as well as small storage capacity and low computational complexity of a client, and has great practical value and broad application prospect.

The invention belongs to the field of computer vision, particularly relates to a development method and system of a small sample classification model based on a graph convolutional neural network, andaims to solve the problems that an existing model is poor in performance and needs a large number of training samples with labels when applied to a new task. The method comprises the steps of extracting a knowledge graph formed by an undirected graph, and obtaining knowledge information related to a task; extracting experience information of the original model in the training process of the old task; fusing the knowledge information and the empirical information to form a new fusion graph; and obtaining a new task-oriented cognitive developed classification model by establishing a relationship between a new task category and an old task category and carrying out model training. According to the method, in the task migration process, a new task can achieve rapid and accurate visual migration without providing a large number of samples with labels, the utilization rate of the model is greatly increased, the cost of training the model is reduced, and the time of training the model is shortened.

Disclosed is an epsilon-greed based online sequential perceiving and opportunity accessing method. In each time slot, users perceive channels sequentially and get access opportunistically to realize transmission. The method includes a step of initializing relative parameters and making access decisions based on online learning in each time slot, and has capabilities of learning environments actively and adapting to dynamic changes of environments. In addition, the method is an online decision implementing method, and the next decision is adjusted in real time according to each decision and feedback by systems, so that long-term accumulated throughput gain of the systems is maximized.

Provided in an embodiment of the present invention is an image encoding method, comprising: determining an intra prediction mode of an encoded block, and constructing a first prediction value of the encoded block according to the intra prediction mode; determining a filtering parameter according to the original value of the coding block and the first prediction value, wherein the filtering parameter comprises a filtering indication parameter; if the filtering indication parameter indicates to perform filtering processing on the first prediction value, performing filtering processing on the first prediction value to obtain an intra-frame prediction value; calculating a prediction difference parameter according to the difference between the original value of the coding block and the intra-frame prediction value; and encoding the intra-frame prediction mode, the filtering parameter and the prediction difference parameter, and writing encoding bits into a code stream. The invention furtherprovides an image decoding method, an encoder, a decoder and a computer storage medium.

The invention discloses a node number variable underwater acoustic network medium access control method based on Q learning, and relates to an underwater acoustic network. The method comprises the following steps: dividing a transmission process of collecting data sensed by underwater acoustic sensor nodes by aggregation nodes into a plurality of time slots, applying a Q learning algorithm, combining feedback signals of the aggregation nodes and the quantity change condition of the sensor nodes, reasonably setting an award mechanism, and carrying out integral award sub-matrix design on a whole line (namely, a sub-matrix) of a Q matrix, not updating the Q matrix element by element, and reasonably distributing the time slot to each sensor node, so the data is not influenced by other sensor nodes in the transmission process, and the data collection conflict of the aggregation node is avoided. The method provided by the invention has the characteristics of high learning speed, high throughput, low energy consumption and strong anti-interference capability, can solve the problem of time slot redundancy when nodes are reduced or the problem of insufficient time slot when nodes are increased due to node death or position drift, and ensures the success rate of underwater acoustic data transmission and high throughput of an underwater acoustic network.

The invention discloses a method and a system for reordering and inversely reordering predicted residual blocks. The method comprises that: according to the amplitude distribution of the predicted residual blocks obtained by statistics, methods for ordering the predicted residual blocks are set, and all possible methods for ordering the predicted residual are two or more than two; and according to the methods for ordering the residual blocks, residuals in the residual blocks are reordered before the residual blocks are converted. By reordering the residual sampling values in the residual blocks in a null field, statistical characteristics of residuals in different modes are approximately same, so the performance approximate to mode related transformation is realized by only using a small number of transformation functions, and the storage complexity of a coder decoder is simultaneously reduced.

This invention discloses a quasi-energy conservation and transformation method and the device used in video or image compact by applying two-dimension transformation of not accurate energy conservation, data of different positions adopt a same orthonormalization process to reduce computation complexity and storage complexity for a code/decode device, which is not only used in video or image code/decode field but also used in other signal processes.

The invention discloses a cloud storage data efficient deduplication auditing method, which belongs to the technical field of cloud storage, and comprises the following steps: S1, establishing a cloud storage system, generating parameters of the cloud storage system, and initializing various parameters of the cloud storage system; according to the method, the convergence key is encrypted by using the idea of blind signature and uploaded to the cloud server, the storage and management problems of the convergence key are solved, the data uploaded by the user is subjected to de-duplication operation by using file-level and file-block-level de-duplication modes, the de-duplication granularity is lower, and the de-duplication efficiency is improved. Meanwhile, a file management directory is introduced to manage contents of files and file blocks, convergence keys and legal users, and the efficiency of data deduplication label query is improved by using a Bloom filter.

Embodiments can provide a coding method, a coding apparatus, and a communications apparatus. The method includes: determining N to-be-coded bits, where N is a positive integer; obtaining a first sequence that includes N polar channel sequence numbers; determining a location of an information bit in the N to-be-coded bits based on the first sequence; and performing polar coding on the N to-be-coded bits to obtain coded bits. The location of the information bit in the to-be-coded bits is determined based on the obtained first sequence that includes N polar channel sequence numbers, so that performance of a polar code can be improved.

The invention relates to an RC-LDPC code construction method based on matrix expansion and a Fibonacci sequence. According to the method, check bits are expanded on the basis of a mother code matrix to obtain a group of RC-LDPC codes with the same information bit length and different code rates. The shift matrix of a mother code matrix and the extension matrix is constructed by utilizing a Fibonacci sequence with special properties, and a final check matrix is obtained after the shift matrix is combined with a sparse matrix. The construction method is simple in construction and low in calculation complexity. Simulation results show that the constructed F-RC-LDPC code has good waterfall zone performance in a large code rate range and has no obvious error floor. Moreover, when the bit errorrate is 10<-6>, the net coding gain of the constructed F-RC-LDPC code is improved to a certain extent compared with other LDPC code patterns with the same code length and code rate. Therefore, the scheme has a practical application value in a communication system.

The invention discloses a dynamic BP decoding algorithm for an LDPC code and provides a local dynamic updating decoding method LRBP (Local Residual BP) based on the LDPC code. According to the method, through local residual based message updating, the BP decoding performance comprising little iteration times is effectively improved, and the method surpasses other dynamic BP decoding algorithms. Different from the traditional dynamic BP decoding method, the LRBP method considers that residuals have time validity, so close attention only paid to the time close residuals, a next to-be-updated message is determined according to the time close residuals, the advantage to to this is that the message updating is always carried out through utilization of the newest message, the convergence of the algorithm can be influenced, and the convergence speed of the BP decoding algorithm is effectively improved.

Aiming at the dynamic BP decoding algorithm of the LDPC code, the invention provides a local dynamic update decoding method (LRBP) based on the LDPC code, and the method adopts the message update based on the local residual error, thereby effectively improving the BP decoding performance within a small number of iterations, and surpassing other dynamic BP decoding algorithms. Different from the conventional dynamic BP decoding method, the LRBP method considers that the residual error has timeliness, so that only the nearest residual error in time is paid attention to, and the next message to be updated is determined. The method has the advantages that the latest message can be always used for updating the message, the convergence of the algorithm can be influenced, and the convergence speed of the BP decoding algorithm is effectively improved.

The invention discloses a data storage method and device, equipment and a storage medium. The method comprises the steps of collecting data from a target data source; determining a field type mapped by the field information according to the field information in the data; and storing the data into a storage space corresponding to the field type in a database table corresponding to the target data source. By setting different field types for the storage space of the database table, a unified storage structure is provided for the data sources with different data structures, the situation that different data storage structures are set for the data sources with different data structures is avoided, the storage complexity of different target data sources is reduced. The data storage steps are simplified, the data storage efficiency is improved, the target data sources of different data structures can reuse the storage logic, and unified storage of the data of the target data sources of different data structures is facilitated.