15results about How to "Implement decoding" patented technology

The invention discloses a drug traceability code acquisition method, system and device based on image recognition, and a medium. The method comprises the steps of obtaining a drug package image; preprocessing the medicine packaging image, and performing edge detection, region segmentation, image correction and barcode analysis on the preprocessed medicine packaging image to obtain medicine traceability code information; obtaining prescription data, and checking the drug traceability code information according to the prescription data; and when the drug traceability code information is checked to be correct, carrying out associative storage on the drug traceability code information and the prescription data. According to the invention, drug traceability code information can be rapidly and accurately identified, and environments such as complex backgrounds, illumination changes, traceability code deformation and the like can be adapted; meanwhile, multiple drug traceability codes can be collected, the purpose of batch collection is achieved, supervision of a drug dispensing end is achieved, medication safety of patients is ensured, drug management can be perfected, and subsequent drug traceability is facilitated.

The invention discloses a scene text recognition method and system for multi-granularity training based on global vectors, and relates to the technical field of scene text recognition. The method comprises the following steps: acquiring a to-be-recognized image, and cutting the to-be-recognized image to obtain a part containing a text; the scene text recognition model is used for carrying out scene recognition on the part containing the text to obtain a recognition result, and training is carried out on the scene text recognition model stage by stage and comprises the steps that training samples are preprocessed, the preprocessed samples are encoded through an encoder, global vectors containing multi-granularity information are extracted, and the global vectors containing the multi-granularity information are obtained; and decoding the global vector by using a decoder to obtain an identification result. According to the method, the information of different granularities of the image can be better utilized, the information of the image can be fully learned from easy to difficult, and a better effect can also be achieved for complex texts and curved texts.

The invention discloses an emotion decoding method based on stereoscopic electroencephalogram signals, and relates to neuroscience and artificial intelligence. Performing data preprocessing on the original electroencephalogram data, and removing irrelevant components through frequency band filtering and filtering noise reduction; dividing continuous SEEG data into discrete time windows, constructing a two-dimensional data graph in a time window form, and implementing data enhancement by using overlapped time windows; time features of each electrode channel are extracted through time convolution, and data dimension characterization is improved; the importance of each channel and position in the feature map is adaptively adjusted by combining a channel and a space attention module, so that the model pays more attention to a key area; fully supervised training is carried out on the emotion decoding network by using the constructed data set, and the relationship between time window data and emotion is learned; and performing emotion prediction on test data by using the trained model, and calculating each decoding index. The method improves the accuracy and speed of emotion decoding, and has a wide application prospect in the fields of emotion recognition, brain-computer interfaces and the like.

The invention discloses an electric vehicle charging load prediction method, system and device and a storage medium, and relates to the technical field of electric power, and the method comprises the steps: generating graph structure charging load data based on multi-station charging load data; based on the graph structure charging load data, executing a graph convolutional network-gating time convolutional network adopting an adaptive adjacency matrix, and generating hidden layer representation containing spatial adjacency information; the hidden layer representation and the external features are spliced and input to a time domain fusion Transform, and an electric vehicle charging load prediction result is output; according to the method, the spatial correlation extraction and time-dependent modeling of the charging load are realized by constructing the graph convolutional network-gating time convolutional network based on the adaptive adjacency matrix and the time domain fusion Transform model, and the precision and interpretability of multi-station load prediction are improved in combination with variable selection and an interpretable multi-head attention mechanism.

The invention discloses a real-time video playing method in a QT environment. The method comprises the following steps: S1, creating a compiling environment; s2, formulating a transmission service rule; s3, an ffmpeg pull stream and a decoding resource are initialized; s4, creating a pull stream main thread; s5, decoding the data packet object obtained by stream pulling; s6, carrying out format conversion on the decoded image frame; s7, obtaining the image frame after format conversion and displaying the image frame in real time; and S8, releasing decoding channel resources. The soft decoding method is adopted, the decoding time is effectively shortened, the RTSP video transmission protocol is adopted to obtain the video stream, the playing delay is effectively shortened, and the video playing fluency is guaranteed.

The present application belongs to the field of defect detection, in particular to a pipe network weld flaw detection defect recognition method and system, comprising: acquiring a pipe network weld initial radiographic image, extracting a global feature atlas through a first deep convolution network, and decoding to generate a defect prediction confidence map; extracting a weld key geometric structure, and constructing a geometric prior weight map; applying Bayesian variational inference to the network, statistically dispersing the results of multiple random forward propagation, representing cognitive uncertainty and generating an uncertainty map; pixel-level weighted fusion of the confidence map, the uncertainty map and the geometric prior weight map to obtain a probability heat map, based on which a composite sampling guide vector field is constructed, sampling points are arranged along the vector field, and a multi-angle scanning imaging system is controlled to collect high-resolution local projection data; three-dimensional reconstruction of the projection data to obtain local features, fusion of the local features and global features through a cross-attention module to generate an enhanced defect representation, and output of the class, three-dimensional spatial position and size of the weld defect.

The utility model discloses a novel double-screen KVM (Keyboard Video Mouse) transceiver, which comprises a KVM transmitter and a KVM receiver, the KVM transmitter comprises a first machine body, a first SoC chip, an audio analog-to-digital conversion chip and two HDMI receiving chips, wherein the first SoC chip, the audio analog-to-digital conversion chip and the two HDMI receiving chips are installed in the first machine body. The KVM receiver comprises a second machine body, a second SoC chip, an audio digital-to-analog conversion chip and two HDMI sending chips, wherein the second SoC chip, the audio digital-to-analog conversion chip and the two HDMI sending chips are installed in the second machine body. The novel double-screen KVM transceiver can support transmission of two paths of HDMI video signals at the same time, one sending terminal can be connected with HDMI signals of double display cards of a computer, one receiving terminal can be connected with two display screens, cost is reduced, and the double-screen display function is achieved. Therefore, a user can connect the two displays and check and operate the content on the remote computer at the same time, and the working efficiency is improved.

A programmable 3D beam shaping device based on beamlet printing technology, comprising a laser source, the laser source emits initial laser, the initial laser adjusts the polarization state through a rotatable half-wave plate, and then is separated into orthogonal s component beam and p component beam through a first polarization beam splitter prism; the p component beam is expanded and then enters a Dammann grating to generate a diffraction beam containing several diffraction orders; the diffraction beam passes through a second polarization beam splitter prism, a quarter-wave plate and a first focusing element in turn, introduces focal shift by a multi-stage phase plate array, adjusts the angle and position offset of the sub-group of the sub-beam array beam by a spatial light modulator to generate a working beam, the working beam returns to the second polarization beam splitter prism from the original light path and is emitted into a working area by a second focusing element, generating a three-dimensional light field; the device can accurately control the three-dimensional irradiance distribution in the processing area, thereby realizing multi-level and depth-adjustable energy deposition, and significantly improving the quality, efficiency and consistency of processing.

The present invention discloses a wavefront sensing system and a wavefront information decoding method using single-channel polarization encoding. The polarization information in the wavefront signal is utilized for the first time to realize polarization encoding and decoding of the spot pattern position information, getting rid of the dependence of the traditional Shack-Hartmann sensor on the photoelectric array sensor, while also solving the pain points in the wavefront detection of dim targets, exploring the application potential of single-point detectors in the field of wavefront sensing, and further expanding the application field of adaptive optical systems.

The present application relates to an animation data encoding and decoding method, device, storage medium and computer equipment, wherein the animation data encoding method comprises: obtaining animation data corresponding to each animation tag code from an animation engineering file; when an attribute type exists in an attribute structure table corresponding to the animation tag code, determining the attribute identification information corresponding to each attribute; encoding the attribute value corresponding to each attribute in the animation data according to the attribute identification information to obtain the attribute content corresponding to each attribute; according to the attribute sorting corresponding to each attribute in the attribute structure table, storing the attribute identification information and attribute content corresponding to each attribute in sequence to obtain the dynamic attribute data block corresponding to the animation tag code. The solution provided by the present application can significantly reduce the file size of animation files.

The application discloses a real-time video playing method in QT environment, which comprises the following steps: S1, creating a compiling environment; S2, making a transmission service rule; S3, initializing ffmpeg stream pulling and decoding resources; S4, creating a stream pulling main thread; S5, decoding a data packet object obtained by stream pulling; S6, converting a decoded image frame; S7, obtaining the converted image frame and displaying in real time; and S8, releasing decoding channel resources. The application adopts a soft decoding method, effectively reduces decoding time, simultaneously adopts an RTSP video transmission protocol to obtain a video stream, effectively reduces playing delay, and guarantees video playing fluency.

This invention belongs to the interdisciplinary field of neuroscience, sensory science and technology and food science. It provides a method for objectively quantifying taste perception using electroencephalography. The invention aims to separate and decode the neural dynamics of the brain in processing taste stimuli of different intensities, and to objectively evaluate and classify them based on individual differences in innate taste sensitivity.

The invention discloses a decoding method for solving overtime memory overflow. The decoding method comprises the following steps: S1, transplanting ffmpeg to an acceleration module platform; s2, setting an initialization resource calling interface; s3, using an ffmpeg dynamic library to obtain a complete compressed image data packet containing a parameter frame; s4, performing hard decoding based on an acceleration module; s5, setting an error callback function to check a decoding state; s6, returning the decoded image to the upper application through an interface function; and S7, releasing decoding channel resources. A soft decoding method is adopted to form a complete data packet containing an image frame and a parameter frame, and then a hard decoding module of a chip is used to complete the decoding process, so that the decoding time is effectively shortened.

The invention belongs to the technical field of data link communication, and particularly discloses a Turbo coding device and decoding device based on an FPGA, and the FPGA of the Turbo coding device is provided with a main control module, a first memory, an interleaving module, a first RSC coding module, a second RSC coding module and a puncturing multiplexing module. A sequence separation module, an iterative decoding module and a main control module are configured on an FPGA (Field Programmable Gate Array) of the Turbo decoding equipment. According to the invention, turbo coding and decoding can be realized based on the FPGA, and by adopting the turbo coding and decoding equipment provided by the invention, the information processing capability of the system can be improved, and meanwhile, the system can be ensured to obtain better bit error rate performance; according to the method, the structure is simple, hardware description language Verilog language can be used, a hardware environment can be built for implementation, and the practicability is high.

The invention relates to the technical field of RFID tag coding, in particular to an RFID tag coding method. Firstly, an RFID tag is embedded in a clothing tag, then when pictures or complex data need to be stored, a cloud box encodes the data, compresses the pictures and converts the pictures into a format suitable for the storage of the RFID tag, then based on a bidirectional authentication method of the RFID tag and the cloud box, to-be-written tag data is written into the RFID tag, then the RFID tag is scanned through a code scanning device, and the data are sent to the cloud box. And on the basis of a readability authentication method between the code scanning equipment and the RFID tag, it is confirmed that the code scanning equipment has a data reading permission, tag data in the code scanning equipment is obtained, the data is transmitted to the cloud box to be decoded, and the cloud box restores coded information into picture or text information. According to the invention, coding, storage and decoding of picture and text data are realized, the intelligent and automatic level of clothing management is greatly improved, and commodity tracking and inventory management are facilitated.