94 results about "Volume (compression)" patented technology

The invention discloses a video stream processing method for dynamic Gaussian compression and adaptive code rate regulation, which is suitable for scenes such as virtual reality, augmented reality and three-dimensional video, and comprises the following steps: S1, Gaussian attribute modeling and initialization; s2, constructing a binary hash grid; s3, constructing a deformation prediction network; s4, designing a mask pruning mechanism; s5, entropy modeling and arithmetic coding and decoding module design; s6, model training; and S7, video stream transmission under multiple code rates. According to the method, a unified scheme combining Gaussian volume cloud coding and adaptive video transmission is proposed for the first time, the video data storage and transmission cost is remarkably reduced, and the comprehensive performance superior to that of an existing method is obtained on multiple real and synthetic data sets.

The invention discloses a neural radiation field rendering method based on dynamic hash coding, and the method comprises the steps: employing the feature sequence data as the input, calculating the density value and color value of each sampling point through the forward propagation of a neural network, carrying out the volume rendering integral operation according to the ray tracing principle in the direction of a ray, and obtaining the feature sequence data; judging a final color output result of the current pixel point; according to an error value between the color output result and a real image, updating a network parameter weight through a back propagation algorithm, and if the error value is greater than a convergence threshold, continuing to iterate the training process to adjust a feature coding strategy to obtain an optimized neural radiation field model parameter; and after the rendering performance configuration parameters are obtained, optimizing a storage allocation strategy of feature data through a memory pool management mechanism, and if the current memory occupancy rate exceeds a safety threshold, starting a data compression algorithm to reduce the storage space requirement, and obtaining a real-time rendering output result. According to the invention, high-quality real-time rendering of the dynamic scene is realized.

The invention relates to the field of traffic storage optimization and traffic data compression, in particular to a multithreading-combined lossless compression method for fine-grained traffic, which comprises the following steps of: acquiring a traffic data file, extracting a header triple based on the traffic data file, generating an identifier according to the header triple, and transmitting the identifier to a server; aggregating the traffic data files with the same identifier to obtain data streams, and generating a sorted data stream size table according to the data volume of the data streams; creating threads according to system computing power, distributing the data flow to the thread with the minimum thread load according to the data flow size table and the thread load condition, and outputting a data flow distribution scheme; and performing fine-grained characterization and serialization on each data stream to obtain an integer sequence and a byte sequence, selecting a compression processing method according to redundancy characteristics of the integer sequence and the byte sequence, obtaining a redundancy-eliminated sequence, and writing the redundancy-eliminated sequence into a compressed file. The invention aims to provide high compression efficiency to reduce the storage cost while ensuring the data precision.

The invention discloses a digital textbook editing and transmitting system with a multilayer architecture. The digital textbook editing and transmitting system comprises a virtual node modeling module, a lightweight storage module, a function editing module, a resource ecological module and a client module, the virtual node modeling is as follows: Slate is used to construct a virtual node model containing elements such as texts, pictures, formulas, interaction components and the like, and each node carries content attributes, hierarchical relationships and interaction instructions; the lightweight storage module is used for converting a virtual node into a JSON format through global-jsdom after a user completes editing, and classifying binary files in the node into resource nodes in an editor. The invention relates to the technical field of data transmission, adopts an AES encryption transmission technology, avoids the data leakage risk caused by plaintext transmission, and improves the data transmission efficiency. And the digital asset security of the publisher and the user is effectively protected. Through node compression storage (JSON format), the server side can classify and store the JSON data, the data volume is greatly reduced, the storage space multiplexing is improved, and the transmission efficiency is improved.

The invention relates to the technical field of machine learning, and discloses a feature disassembly and compression method and device, equipment and a storage medium, and the method comprises the steps: obtaining an initial feature of to-be-predicted image data, determining an expert activation weight corresponding to a downstream prediction task based on a hybrid expert network, the initial features are processed through a plurality of expert networks in the hybrid expert network to generate disassembled features, the plurality of expert networks are networks constrained through a low-rank matrix, the disassembled features are quantized to generate quantized features, entropy coding is carried out on the quantized features through a hyper-prior network, and compressed features are obtained. According to the method, the expert activation weight is dynamically adjusted along with task requirements, different task scenes are adapted, initial features are disassembled based on a plurality of expert networks, task irrelevant information is explicitly stripped, task key features are reserved, redundant data volume is reduced, quantized coding is performed on disassembled feature input, code stream self-adaption is realized, and bandwidth occupation is reduced.

The embodiment of the invention provides a tag data processing method, RFID middleware and a storage medium. The method comprises the following steps: receiving and caching tag data of a plurality of tags; compressing the label data according to a predetermined compression mode to obtain compressed data; wherein the compression mode is determined according to a data compression level and a label inventory rate; and sending the compressed data. According to the embodiment of the invention, the data processing efficiency is improved, the equipment resource occupancy rate, the data traffic and the bandwidth occupancy of network transmission of middleware at a high inventory rate are greatly reduced, and meanwhile, the mass data processing pressure of an upper platform is reduced. According to the embodiment of the invention, the platform side can receive total label data, and algorithm application development is facilitated. In addition, according to the embodiment of the invention, the packet number and the data compression level are adjusted according to different service requirements, and the method is suitable for service scene requirements with different data volume and time delay sensitivity.

The invention belongs to the technical field related to volume video coding compression, and particularly relates to a multi-view video compression method and application thereof, and the method comprises the steps: initializing the information of each anchor point according to a video first frame; performing quantization and entropy coding on the current anchor point information of the partial anchor points by adopting a preset quantization offset parameter, and adding coding results to obtain Loss2; respectively adding corresponding quantitative offsets to current anchor point information of all anchor points, respectively and correspondingly inputting static and dynamic characteristics with offsets of each anchor point into mlp of neural Gaussian static and dynamic attributes to obtain static and dynamic attributes of each neural Gaussian of the anchor point, obtaining deformation attributes from the dynamic attributes, and obtaining deformation attributes from the deformation attributes; rendering based on the attribute information of each neural Gaussian to obtain a picture, and comparing the picture with a real picture to obtain Loss1; the mlp and all anchor point information are optimized through the two Loss, after optimization, the optimized anchor point information is quantized through Hash coding, entropy coding is conducted, and the coding result and the optimized mlp are compression results. The compression rate can be greatly improved.

The invention relates to the technical field of oral images, and discloses an oral panoramic image real-time compression transmission and 3D reconstruction cloud system. The system comprises an image acquisition module, a compression processing module, a cloud transmission module and a 3D reconstruction module. The image acquisition module acquires initial image data and pre-processes the initial image data to obtain target image data; the compression processing module generates a compression strategy through block coding, and when conditions are met, sub-compression units are divided, and dynamic bit allocation is optimized; the cloud transmission module extracts the compressed code streams, constructs real-time and delayed transmission queues in a classified manner, ranks the queues, verifies the integrity of the code streams to be verified and counts the complete number; and the 3D reconstruction module determines an effective data size according to the complete number and the compressed code streams which do not pass the verification, and starts a cloud reconstruction engine. The system improves the oral image processing efficiency and the 3D reconstruction quality through dynamic compression, intelligent transmission and precise verification, and is suitable for oral medical diagnosis scenes.

The invention discloses a quantum computing gate circuit expression and compression method and a medium, and relates to the technical field of quantum computing, and the method comprises the steps: setting a conversion expression character between a user side and a quantum computing cloud platform, or setting the conversion expression character between the quantum computing cloud platform and a terminal quantum computer for unifying quantum languages; the conversion expression character is divided into three segments which are an operation code, a bit code and a parameter code in sequence, the operation code refers to quantum gate operation, the bit code refers to a quantum bit number, and the parameter code refers to a floating-point number related to the operation code; the core of the method is that the file volume is compressed and the quantum language is unified, so that the problem of transmission efficiency of existing codes among different terminals is solved.

The invention discloses an image compression method based on adaptive quadtree segmentation and standard deviation optimization, and provides an efficient compression scheme driven by dynamic statistics aiming at the problems of over-segmentation, low calculation efficiency and detail loss under a high compression rate in a traditional quadtree compression algorithm. Specifically, the compression ratio and the reconstruction quality are effectively balanced by calculating the RGB channel standard deviation of an image region in real time and combining region area weighted income analysis to adaptively control the recursive segmentation depth; an area weighted error constraint mechanism is introduced, segmentation of a large-size low-variance region is terminated preferentially, and the redundant data volume is remarkably reduced; a standard deviation-driven priority ranking strategy is adopted, fine processing is performed on high-complexity sub-regions, and rapid approximation is realized by combining mean color filling, so that the algorithm efficiency is optimized while the peak signal-to-noise ratio is ensured. Through the synergistic effect of the technologies, on the premise that high-frequency details of the image are reserved, calculation complexity and storage requirements are greatly reduced, the method is suitable for real-time image transmission, mobile terminal processing and low-bandwidth communication scenes, and a reliable solution is provided for high-precision and low-delay compression requirements.

This invention discloses an efficient RNA data preprocessing method to address the problems of low processing efficiency and low information density in high-throughput sequencing data. Its core steps include: (1) introducing a parallel processing scheme for high-throughput sequence data, rapidly mapping RNA-seq data to a reference genome to generate a BAM file; (2) extracting base sequences and expression levels and storing them as compact PKL format files; (3) extracting all exon position information by parsing the genome annotation file; (4) combining multi-sample expression level data to screen for highly variable exons and constructing a high-information-density feature list based on the sample set; and (5) accurately extracting target sequences from the preprocessed file based on this list. Compared to traditional methods, this innovative approach achieves triple optimization: full-process parallel processing for accelerated computation, high-compression data storage, and adaptive feature selection. Processing speed is increased by 3-5 times, and data volume is reduced by more than 90%, making it suitable for high-throughput RNA-seq data analysis with large sample sizes.

The present application relates to a compression method for zero-trust network access technology. The method includes: when there is data update in the system, judging the type of data to be updated to obtain a judgment result; when the updated data in the system is a file, when the updated file meets the preset access volume condition, triggering a compression mechanism to compress the updated file; when the updated data in the system is a policy or a user, calculating the correlation degree between the updated policy or user and the file, and judging whether to trigger the compression mechanism to compress the file corresponding to the updated policy or user according to the correlation degree. The present application can identify the type of updated data, apply corresponding judgment criteria, obtain a reasonable judgment result for the updated data, intelligently compress and cache the data that meets the criteria, and save as much network bandwidth resource as possible with as little computing power as possible.

The invention provides a data storage and reading method, device and equipment, and the data storage method comprises the steps: extracting the features of to-be-stored data, and applying for an index storage position in a traversed current data volume according to a level sequence from a higher level to a lower level, when it is determined that first to-be-stored data which does not meet the data similar compression condition exists, the first to-be-stored data is written into an applied target storage position, and the target storage position is written into an index storage position; when it is determined that second to-be-stored data meeting the data similar compression condition exists, the second to-be-stored data is compressed, the to-be-stored data is updated, the index storage position applied by the to-be-stored data in the next-level data volume is written into the current data volume, and if the level of the current data volume is the lowest level, the next-level data volume is created, and a data storage instruction is triggered. Through the method, the to-be-stored data is stored after being subjected to multi-stage compression based on the multi-stage data volume, so that the compression rate of the data is improved.

The present invention relates to the technical field of data processing, and specifically provides a method and system for compressing and reconstructing Internet of Things data based on deep compressive sensing. The collected data of several sensors is used as the original data, and compressive sensing is performed to obtain compressed data; the compressed data is decoded and reconstructed; the compressed data is mapped to the original space to obtain initialization data, which is input into the residual decoder to obtain decoded data; the decoded data of the residual decoder is mapped to a high-dimensional embedding space for data reconstruction to obtain output data; the output data is mapped to the original space through a fully connected layer to obtain reconstructed data, and the reconstructed data is the original data. A residual sensing block is used to adaptively generate a learnable measurement matrix, reducing the data transmission volume in the Internet of Things and improving the data reconstruction accuracy. The residual decoder and the data reconstruction decoder are combined to complete the reconstruction of Internet of Things data by extracting key features from the compressed data.

Embodiments of the present application provide a BIM model value-driven three-dimensional unit compression method and system, which comprises: obtaining an original BIM model containing geometric information and material information and a general compression target; and calculating a component value score of each component in the original BIM model; based on the original BIM model and the component value score, performing voxelization processing and compression unit division to obtain a plurality of compression units; independently compressing and adaptively entropy encoding the plurality of compression units respectively to generate code streams corresponding to the plurality of compression units respectively; collecting the code streams corresponding to the plurality of compression units respectively to obtain a total volume of the plurality of compression units; based on the total volume and the general compression target, determining an encoding parameter to complete compression and obtain final code streams corresponding to the plurality of compression units respectively; and storing the final code streams to obtain a final compression file. The above scheme can accurately balance the compression ratio and data fidelity, and improve the encoding and decoding efficiency.

The invention discloses a reverse debugging increment recording method based on dynamic layering and predictive filtering, which comprises the following steps of: analyzing a binary file division region of a target program, establishing a global region mapping table among an address range, a region type and debugging attributes, executing the binary file by a reverse debugger, constructing a first mapping table for modifying a record circular queue, according to the global region mapping table, performing differentiation recording on the region to form a first record, a stack frame snapshot, a second record and a register block snapshot, and performing repeated coverage filtering to reduce the recorded data volume to form an execution state track; and performing intelligent compression on stack frame snapshots, heap memories, global variable records, register snapshots and variable values in the execution state trajectory, and finally forming a low-overhead and accurate execution state trajectory to support reverse debugging backtracking.

The present invention belongs to the field of data processing technology, and specifically relates to a data information processing method and system based on big data. The method includes: constructing a Huffman tree based on frequency statistics of all data in the data information; obtaining a special layer in which the increase in data volume is less than the decrease in data volume when the path code from the root node to the branch node is replaced with a fixed-length code of the sequence number of the branch node in the layer; and subsequently encoding all data through the Huffman tree and the special layer. For target data corresponding to the special layer, the branch node of the leaf node corresponding to the target data in the special layer is used as a guide node, and the fixed-length code of the guide node in the special layer is spliced with the path code from the guide node to the leaf node corresponding to the target data to obtain the encoding result of the target data. The present invention improves the compression rate of the data information compression result.

Three-dimensional sensing data is transferred efficiently by applying compressed sensing using dictionary learning. Conventional image-based lossy compression improves a feature according to which power is concentrated on low-frequency components and deterioration of three-dimensional information increases. By changing the decimation rates for the depth information and the color information and using a dictionary vector created from the other restoration result for restoration, the original result is reconstructed using a small data volume.

The application relates to the technical field of CAE preprocessing and cloud cooperation, and discloses a lightweight CAE geometry engine and a cloud real-time cooperation method and system. A cloud server acquires an original CAE geometry model, performs topological simplification and data compression on the original CAE geometry model, and generates a lightweight geometry data file containing multiple resolution levels; the cloud server pushes the received difference data packet to other online cooperation terminals; the other cooperation terminals parse the unique version number carried by the difference data packet through a locally loaded WebAssembly geometry engine module, and perform operation replay updating on the local geometry model according to the differential data contained in the difference data packet. According to the scheme, the original CAE model data volume is reduced by 90%-95%, the WebAssembly engine realizes a 100,000-face model rendering frame rate of 30-60 fps, and the operation response delay is less than or equal to 100 ms, thereby breaking through the performance bottleneck of browser-side geometry processing.

The invention provides a real-time three-dimensional reconstruction method and system based on unmanned equipment. In order to solve the lag problem that in traditional reconstruction, collection needs to be conducted firstly and then reconstruction needs to be conducted, and the composite bottleneck that airborne computing power is limited and weak network bandwidth is insufficient, a unified time signal is generated through a hardware synchronization module, and microsecond-level synchronization of multi-source data is achieved; aiming at heterogeneous data characteristics, the airborne end respectively constructs batch processing assembly lines of point cloud octree compression, image JPEG coding and IMU byte alignment, so that the data volume is remarkably compressed while the computing power overhead is reduced; a QoS mechanism and local persistent cache of an MQTT protocol are introduced into a transmission layer, and reliable arrival of data during network fluctuation is guaranteed through a breakpoint resume technology. In conclusion, end-to-end real-time processing of software and hardware collaboration is realized, bandwidth occupation is effectively reduced, and stable and high-precision real-time three-dimensional reconstruction is realized in a weak network environment.

The invention relates to the technical field of computers, in particular to a picture custom cutting and volume compression method and system based on a Web end. The method comprises the following steps: acquiring a picture file, and loading and initializing the picture file to obtain a processed picture; performing incremental rendering on the processed picture based on a dynamic center anchor point scaling algorithm to obtain a rendered picture; selecting and intercepting a picture pixel data area based on the rendered picture to obtain a picture file stream of a lightweight screenshot area; and carrying out image dynamic quality compression and output on the picture file stream of the lightweight screenshot area to complete picture custom cutting and volume compression based on the Web end. According to the method, the problem that the size of the picture can be reduced by cutting the picture or reducing the resolution of the picture when the picture file is applied to the function requirements of uploading and downloading the picture file is solved, the problem that huge bandwidth resources are occupied in network transmission is solved, and the occupied storage space of a server is optimized and reduced.

This invention provides a data compression method, apparatus, electronic device, and storage medium. The method includes: segmenting the original data using a breakpoint detection algorithm; fitting each segment of data using a fitting algorithm; and obtaining the restored data using a decompression algorithm. This addresses the problem of excessively large data volumes and high storage and transmission costs in current massive data acquisition scenarios, thus promoting the further development of various data platforms for digital transformation in the big data era. The method of segmenting the original data using a breakpoint detection algorithm and fitting each segment of data using a fitting algorithm effectively achieves efficient data compression. The restoration of the data through a decompression algorithm further addresses the problem of excessively large data volumes and high storage and transmission costs in current massive data acquisition scenarios, thus promoting the further development of various data platforms for digital transformation in the big data era.

This invention discloses a method, apparatus, and electronic device for effectively adjusting the compression ratio of a reference frame image. The method includes: S1, dividing the reference frame image to be compressed into several n×n blocks; S2, initializing the target bit count of the n×n blocks according to the total data volume of the reference frame image to be compressed and the target compression ratio requirement, and obtaining the target bit count occupied by each pixel; S3, calculating the target bit rate weight parameter λ based on the obtained target bit count occupied by each pixel. real S4, calculate the quantization parameters based on the relationship between the quantization parameters and the target bit rate weight parameters, and perform encoding quantization; S5, complete the encoding process of the current n×n block and obtain the actual number of encoded bits for each pixel of the current n×n block; S6, calculate the actual bit rate weight parameters based on the actual number of encoded bits; S7, update the parameters related to the features of the reference frame image and update the remaining allocable bits, and return to step S2.

The embodiment of the invention provides a BIM (Building Information Modeling) value-driven three-dimensional unitization compression method and system. The method comprises the following steps: acquiring an original BIM containing geometric information and material information and an overall compression target; calculating a component value score of each component in the original BIM model; based on the original BIM model and the component value score, voxelization processing and compression unit division are carried out, and a plurality of compression units are obtained; performing independent compression and adaptive entropy coding on the plurality of compression units to generate code streams corresponding to the plurality of compression units; summarizing the code streams corresponding to the plurality of compression units to obtain the total volume of the plurality of compression units; determining coding parameters based on the total volume and the total compression target, and completing compression to obtain final code streams corresponding to the plurality of compression units; and storing the final code stream to obtain a final compressed file. According to the scheme, the compression ratio and the data fidelity can be accurately balanced, and meanwhile, the coding and decoding efficiency is improved.

The application discloses a data compression method and system for an SSD, and relates to the technical field of SSD data compression. The method comprises write IO path processing and read IO path processing. The write IO path processing comprises the following steps: after receiving host data, processing the data in 4K units, judging compressibility, performing compressed storage according to the remaining space of the cache, synchronously updating metadata, and when the data volume of the cache reaches a threshold, allocating NAND space, encoding and writing the data, and updating the address mapping. The read IO path processing comprises the following steps: receiving a read command, searching for the mapping to obtain a NAND physical address, judging whether the data is compressed through metadata after reading and decoding the data, transmitting the decompressed compressed data and uncompressed data to the cache, and transmitting the data to the host after passing the PI check. The application effectively avoids the problem that the current compression technology causes an increase in meta data, thereby excessively occupying LDPC parity space, leading to an increase in LDPC code rate and causing a decrease in LDPC error correction capability.

A file system volume's space is internally allocated as a set of containers, with each container corresponding to a different subset of storage clusters. The set of containers includes a first container corresponding to a first subset of storage clusters storing first data that is compressed according to a first set of compression attributes, a second container corresponding to a second subset of storage clusters storing second data that is uncompressed, and a third container corresponding to a third subset of storage clusters that are free. Based on determining that the second data of the second container is to be compressed, compressed second data is created based on a second set of compression attributes. The second set of compression attributes is different than the first set of compression attributes. The compressed second data is written to the third subset of storage clusters of the third container.

The invention provides an efficient real estate file cloning method, belongs to the field of real estate data management, and adopts an intelligent algorithm to quickly identify key files in a real estate data management system and intelligently select files needing to be copied according to actual requirements. A parallel processing technology is adopted, and data is read from multiple source positions at the same time. And lossless compression processing is performed on the read data, so that the data transmission quantity is reduced, and the network load and the storage cost are reduced. A high-speed network interface is utilized to ensure that data is transmitted from a source end to a target end, a hash value calculation verification mechanism is adopted in the transmission process, and early warning is carried out in time once data errors or damages are found. Data are written into a storage device at a target end according to a file structure and an organization mode of a source end, and files are accessed to an existing real estate data management system so as to be better managed. According to the invention, the backup and migration requirements of the real estate industry on key files such as mass production data, equipment state information, security monitoring records and the like are met.

The invention discloses a controllable compression ratio image compression method based on JPEG-LS (Joint Photographic Experts Group-Least Squares), relates to the technical field of image processing, and solves the problems that in an existing image compression method, compression parameters are fixed, so that the difference of image compression ratios is large, the data volume of compressed images is difficult to estimate, and resources of rear-end storage equipment and data transmission equipment are wasted. Setting a target compression ratio, an image depth and parameter values of the NEAR adjustment parameter table of the to-be-compressed image; partitioning the to-be-compressed image according to a target line number, and sending the partitioned images into a JPEG-LS encoder one by one for image compression; according to the method, the NEAR value of the JPEG-LS compression module is adjusted in real time by using the difference value between the compression ratio of the current block image and the target compression ratio, so that the final compression ratio of the whole image is stabilized near the target compression ratio, the controllability of the image compression ratio is effectively improved, and the resource waste of rear-end storage equipment and data transmission equipment is reduced.

The invention relates to a video data processing method and system, electronic equipment and a storage medium. The method comprises the following steps: acquiring an original image frame; identifying a target object in the original image frame, and determining a target area where the target object is located; determining a background region based on the target region; acquiring a preset compression parameter; the preset compression parameters comprise a first compression parameter and a second compression parameter; according to a preset compression parameter, coding the target region and the background region to obtain a coded data packet; and generating target video data based on the coded data packet. According to the technical scheme provided by the invention, the target region and the background region are distinguished in the same original image frame, and different compression parameters are adopted for coding, so that the image presentation quality of the target region can be improved while the overall coding resource occupation and the data volume are controlled; and the comprehensive display effect of the video data can be improved in a scene with limited bandwidth or processing capability.

The invention provides a model training method and a document compression method for realizing AI Agent based on AI, RPA and LLM, and relates to the field of AI, RPA, LLM and AI Agent. The method comprises the steps of obtaining a sample document image and a sample original document corresponding to document content displayed by the sample document image; compressing the original sample document by adopting a compression representation grammar to obtain a compressed sample text; and training a visual language model (VLM) based on the sample document image and the sample compressed text, so that the trained VLM learns a compressed representation grammar. Therefore, the trained VLM can effectively filter redundant information corresponding to the meaningless visual elements (such as table null lines and the like), output the compressed text conforming to the compressed representation grammar, reduce the data volume of the generated text, reduce the resource burden of data transmission and storage, do not need subsequent extra redundant information removal operation, and improve the data processing efficiency in subsequent data processing. And the calculation overhead can be reduced, and the model reasoning accuracy and reliability can be improved.