736 results about "Encoder decoder" patented technology

Techniques and tools for encoding and decoding motion vector information for video images are described. For example, a video encoder yields an extended motion vector code by jointly coding, for a set of pixels, a switch code, motion vector information, and a terminal symbol indicating whether subsequent data is encoded for the set of pixels. In another aspect, an encoder / decoder selects motion vector predictors for macroblocks. In another aspect, a video encoder / decoder uses hybrid motion vector prediction. In another aspect, a video encoder / decoder signals a motion vector mode for a predicted image. In another aspect, a video decoder decodes a set of pixels by receiving an extended motion vector code, which reflects joint encoding of motion information together with intra / inter-coding information and a terminal symbol. The decoder determines whether subsequent data exists for the set of pixels based on e.g., the terminal symbol.

Techniques and tools for coding / decoding of video images, and in particular, B-frames, are described. In one aspect, a video encoder / decoder determines a fraction for a current image in a sequence. The fraction represents an estimated temporal distance position for the current image relative to an interval between a reference images for the current image. The video encoder / decoder processes the fraction along with a motion vector for a first reference image, resulting in a representation of motion (e.g., constant or variable velocity motion) in the current image. Other aspects are also described, including intra B-frames, forward and backward buffers for motion vector prediction, bitplane encoding of direct mode prediction information, multiple motion vector resolutions / interpolation filters for B-frames, proactive dropping of B-frames, and signaling of dropped predicted frames.

A deblocking filter used in a video encoder / decoder based on a multilayer is disclosed. The deblocking filtering method includes choosing a present block and its neighboring block on which a deblocking filtering will be performed, judging whether the present block and the neighboring block have been coded by the intra BL mode, choosing a deblocking filter strength according to the result of the judgment, and performing the deblocking filtering with respect to a boundary between the present block and the neighboring block and according to the deblocking filter strength.

Techniques and tools for encoding and decoding predicted images in interlaced video are described. For example, a video encoder or decoder computes a motion vector predictor for a motion vector for a portion (e.g., a block or macroblock) of an interlaced P-field, including selecting between using a same polarity or opposite polarity motion vector predictor for the portion. The encoder / decoder processes the motion vector based at least in part on the motion vector predictor computed for the motion vector. The processing can comprise computing a motion vector differential between the motion vector and the motion vector predictor during encoding and reconstructing the motion vector from a motion vector differential and the motion vector predictor during decoding. The selecting can be based at least in part on a count of opposite polarity motion vectors for a neighborhood around the portion and / or a count of same polarity motion vectors.

The invention discloses a remote sensing image semantic segmentation method based on attention multi-scale feature fusion, and the method comprises the steps: establishing a semantic segmentation network based on attention multi-scale feature fusion, building a training data set, and carrying out the network parameter training through employing the training data set. And performing semantic segmentation on to-be-tested data by using the trained network during testing. And the network is a lightweight encoder-decoder structure. Wherein the idea of an image cascade network is introduced; meanwhile, coding features and decoding features are optimized by using an attention mechanism, a multi-scale attention optimization module, a multi-scale feature fusion module and a boundary enhancement module are constructed, feature maps of different scales are extracted and fused, training is guided by using multi-scale semantic tags and boundary tags, and semantic segmentation of remote sensing images can be effectively carried out.

A system on chip processor for a multimedia device includes: a pre-processing circuit to convert an external image signal into a compressed input signal for compressing; an encoder / decoder circuit to generate compressed data by compressing the compressed input signal and outputting a coded image signal by decompressing the compressed data; a post-processing circuit to convert the coded image signal into a signal that can be used by an image displaying device; a first system bus connected with pre-processing circuit and post-processing circuit; a second system bus connected with the encoder / decoder circuit; a first bridge DMA circuit to mutually transmit data between first system bus and second system bus; and a controller to control the operation of the circuits.

Deblocking filter used in a video encoder / decoder based on a multilayer. In deciding a deblocking filter strength when performing a deblocking filtering with respect to a boundary between a current block coded by an intra-BL mode and its neighboring block, it is determined whether the current block or the neighboring block has coefficients. The filter strength is decided as a first filter strength if it is determined that the current block or the neighboring block has the coefficients, and the filter strength is decided as a second filter strength if it is determined that the current block or the neighboring block does not have the coefficients. The first filter strength is greater than the second filter strength.

The invention provides a method and a system for deep nerve translation based on character encoding. A combined nerve network model is established by using an RNN to cover the whole translation process, and translation tasks are directly completed from the perspective of an encoder-decoder framework. The method comprises the following steps: A, word vector generation: performing word segmentation on character-level input data through neural network modeling and generating a word vector; B, language model generation: establishing grammar rules by utilizing the characteristic of memory of the recurrent neural network in time; C, word alignment model generation: obtaining the probability of translating multiple words in a source language statement into target language words; D, output: translating an inputted source language into a target language; E, translation model combination: establishing a deep nerve translation model (RNN-embed) based on character encoding in combination with neural network models in the four steps and accelerating model training by using CPU parallel computation.

Apparatus and method to decode video data while maintaining a target video quality using an integrated error control system including error detection, resynchronization and error recovery are described. Robust error control can be provided by a joint encoder-decoder functionality including multiple error resilience designs. In one aspect, error recovery may be an end-to-end integrated multi-layer error detection, resynchronization and recovery mechanism designed to achieve reliable error detection and error localization. The error recovery system may include cross-layer interaction of error detection, resynchronization and error recovery subsystems. In another aspect, error handling of a scalable coded bitstream is coordinated across a base-layer and enhancement layer of scalable compressed video.

A prediction encoder / decoder, a prediction encoding / decoding method, and a computer readable recording medium having a program for the prediction encoding / decoding method recorded thereon. The prediction encoder includes a prediction encoding unit that starts prediction at an origin macroblock of an area of interest of a video frame, continues prediction in an outward spiral in the shape of square rings composed of macroblocks surrounding the origin macroblock, and encodes video by performing intra-prediction using information about a macroblock that has just been coded in a square ring including a macroblock to be coded and a macroblock in a previous square ring and adjacent to the macroblock.

An autonomous and portable smartcard reader device incorporates a high level of embedded security countermeasures. Data transfers are encrypted with two specific input devices, namely a light sensor and PIN or other keyboard entry, and at the output through the use of a dual-tone encoder-decoder. The unit may be used alone or as a plug-in to another device such as a PDA, cell phone, or remote control. The reader may further be coupled to various biometric or plug-in devices to achieve at least five levels of authentication, namely, (1) the smartcard itself; (2) the smartcard reader; (2) the PIN; (3) private-key cryptography (PKI); and (5) the (optional) biometric device. These five levels account for an extremely strong authentication applicable to public networking on public / private computers, and even on TV (satellite, cable, DVD, CD AUDIO, software applications. Transactions including payments may be carried out without any risk of communication tampering, authentication misconduct or identity theft. The invention finds utility in a wide range of applications, including commercial transactions and security, including the generation of prepaid credit / debit card numbers, vehicle rental, and other uses.

A bluetooth headset for transmitting and receiving digital data and analog audio signal is disclosed. The bluetooth headset comprises: a bluetooth module for transmitting and receiving a wireless signal comprising a control signal, the format of the wireless signal being compliant with a bluetooth specification; an audio encoder / decoder for encoding an analog audio input signal to generate digital data, and decoding the digital data to generate an analog audio output signal; a memory unit for storing the digital data; and a control unit which is coupled to the bluetooth module for converting the wireless signal from received by the bluetooth module into the digital data, and transferring the digital data to one of the memory unit or the encoder / decoder.

The invention discloses a deep learning-based blog text abstract generation method. The method comprises the following steps of: crawling blog data; preprocessing the crawled blog data and selecting blog text data; converting the selected blog text data into vector matrix data according to a Chinese word vector dictionary; constructing a deep learning encoder-decoder model, separately training an encoder and a decoder of the model, and connecting the encoder and the decoder for use after the training is completed; and repeating the steps S01 to S03 to obtain generated data, and generating a predicted abstract from the generated data through the trained model. According to the method, text abstracts of blogs are automatically generated on the basis of a deep learning frame encoder-decoder, and deeper semantic relation of the blogs can be obtained at the same time. The generated text abstracts can visually display the main content of the current blog, so that the text abstracts have wide application prospect.

In one aspect, an encoder / decoder receives information for four field motion vectors for a macroblock in an interlaced frame-coded, forward-predicted picture and processes the macroblock using the four field motion vectors. In another aspect, an encoder / decoder determines a number of valid candidate motion vectors and calculates a field motion vector predictor. The encoder / decoder does not perform a median operation on the valid candidates if there are less than three of them. In another aspect, an encoder / decoder determines valid candidates, determines field polarities for the valid candidates, and calculates a motion vector predictor based on the field polarities. In another aspect, an encoder / decoder determines one or more valid candidates, determines a field polarity for each individual valid candidate, allocates each individual valid candidate to one of two sets (e.g., opposite polarity and same polarity sets) depending on its field polarity, and calculates a motion vector predictor based on the two sets.

For interlaced B-frames, an encoder / decoder computes direct mode motion vectors for a current macroblock by selecting at most one representative motion vector for each of the top and bottom fields of the co-located macroblock of the previously decoded, temporally subsequent anchor. For example, the selecting is performed based at least in part on the mode of coding the current interlaced B-frame's macroblock (e.g., 1MV mode, 2 Field MV mode, etc.). For interlaced B-fields, an encoder / decoder selects direct mode motion vectors using logic that favors the dominant polarity if the corresponding macroblock in the corresponding field of the next anchor picture was coded using four motion vectors. For example, if the corresponding macroblock's same polarity motion vectors outnumber its opposite polarity motion vectors, the encoder / decoder calculates the median of the same polarity motion vectors to obtain a motion vector for deriving direct mode motion vectors.

A remote interface network in which multiple remote HID encoder / decoder units share a common physical transport medium for connecting to one or more processing unit encoder / decoders is described. In one embodiment, the physical transport medium includes an optical shared media transport network. Each remote HID encoder / decoder unit can support one or more remote HIDs. The processing unit encoder / decoder can support one or more Pus. The network can be used, for example, in office, hospital, dense seat (e.g., aircraft, bus, etc.) and content provider networks.

In a multi-channel encoder generating several different parameter sets for reconstructing a multi-channel output signal using at least one transmission channel, the data stream is written such that the two parameter sets are decodable independently of each other. Thus, a multi-channel decoder is enabled to skip a parameter set which is marked as optional and / or has a higher version number when reading the data stream and still to perform a valid multi-channel reconstruction using a data set marked as mandatory or a data set having a sufficiently low version number. This achieves a flexible encoder / decoder concept suitable for future updates characterized by backward compatibility and reliability.

An encoder / decoder is based on a combination of two audio or video channels to obtain a first combination signal as a mid-signal and a residual signal derivable using a predicted side signal derived from the mid-signal. A decoder uses the prediction residual signal, the first combination signal, a prediction direction indicator and prediction information to derive decoded first channel and second channel signals. A real-to-imaginary transform can be applied for estimating the imaginary part of the spectrum of the first combination signal. The prediction signal used in the derivation of the prediction residual signal, the real-valued first combination signal is multiplied by a real portion of the complex prediction information and the estimated imaginary part of the first combination signal is multiplied by an imaginary portion of the complex prediction information.

A broadband network system configured to transfer user datagram protocol with value-added (UDPVA) packet from to a broadband interface unit (BIntU) transceiver. The broadband network system may comprise a data distribution center and the BIntU transceiver. The BIntU transceiver is configured to generate the UDPVA packet. The BIntU transceiver includes an encoder / decoder (codec) configured to alternatively code or decode UDP frame information and a digital signal processor (DSP) portion. The DSP portion is coupled to the codec. In one aspect, the DSP portion includes a stack that temporarily stores the UDPVA packet within the stack. In one aspect, the UDPVA packet is in a form to be transmitted directly to a network destination address device. In another aspect, the UDPVA packet is available to delivery to a network destination address or storage located on a local area network or a wide area network. In certain aspects, the BIntU transceiver is configured to transmit the UDPVA packet to the data distribution center.

The present invention relates to a method and device for generating a depth image, a method for encoding / decoding the depth image, an encoder / decoder for the same, and a recording medium recording an image generated by the method, which are related to a depth image encoding method that can effectively reduce a bit generation rate using a reference image obtained by at least one camera and improve encoding efficiency. A depth image generating method according to an embodiment of the invention includes a step (a) of obtaining a depth image at a viewpoint and setting the obtained depth image to a reference image; a step (b) of applying a 3D warphing method to the reference image and predicting and generating a depth image at a specific viewpoint; and a step (c) of removing a hole that exists in the predicted and generated depth image.

The invention discloses an aspect-level emotion classification model and method based on dual-memory attention, belonging to the technical field of text emotion classification. The model of the invention mainly comprises three modules: an encoder composed of a standard GRU loop neural network, a GRU loop neural network decoder introducing a feedforward neural network attention layer and a Softmaxclassifier. The model treats input statements as a sequence, based on the attention paid to the position of the aspect-level words in the sentence, Two memory modules are constructed from the originaltext sequence and the hidden layer state of the encoder respectively. The randomly initialized attention distribution is fine-tuned through the attention layer of the feedforward neural network to capture the important emotional features in the sentences, and the encoder-decoder classification model is established based on the learning ability of the GRU loop neural network to the sequence to achieve aspect-level affective classification capabilities. The invention can remarkably improve the robustness of the text emotion classification and improve the classification accuracy.

An encoder, a decoder, corresponding encoding and decoding methods, video signal structure, and storage media are provided for scalable video encoding and decoding. A scalable video encoder includes an encoder (100) for encoding a picture to form a base layer bitstream and an enhancement layer bitstream. The base layer bitstream and the enhancement layer bitstream are formed by partitioning the picture into a plurality of image blocks, grouping the plurality of image blocks into one or more slice groups in the base layer bitstream and into two or more slice groups in the enhancement layer bitstream, encoding all of the one or more slice groups in the base layer bitstream and less than all of the two or more slice groups in the enhancement layer such that at least one slice group from among the two or more slice groups is intentionally un-encoded in the enhancement layer bitstream, encoding a syntax element in a header to indicate the at least one intentionally un-encoded slice group in the enhancement layer.

The invention relates to a base matrix, a encoder / decoder of Low Density Parity Check (LDPC) codes and a generation method thereof. The encoder / decoder is determined uniquely by the parity check matrix of the LDPC codes. With different code sizes, said parity check matrix can be obtained by expanding different base matrixes, and also can be obtained by expanding a unique base matrix after correction. Elements of said base matrix must meet an inequation in which the girth value is up to the standard, e.g. when girth ≧6, for any element i, j, k, l in the matrix which forms the short loop having a length of 4 in anticlockwise, there are always mod(i−j+k−l, z) unequal to 0. By this invention, the girth of the constructed base matrix is made as large as possible, the amount of the shortest loops is as few as possible when it is the same girth, thus the curve of bit error ratio (BER) is dropped quickly, the error floor of the LDPC codes is eliminated effectively, and the optimal performance of the LDPC codes is obtained.

An autonomous and portable smartcard reader device incorporates a high level of embedded security countermeasures. Data transfers are encrypted with two specific input devices, namely a light sensor and PIN or other keyboard entry, and at the output through the use of a dual-tone encoder-decoder. The unit may be used alone or as a plug-in to another device such as a PDA, cell phone, or remote control. The reader may further be coupled to various biometric or plug-in devices to achieve at least five levels of authentication, namely, (1) the smartcard itself; (2) the smartcard reader; (2) the PIN; (3) private-key cryptography (PKI); and (5) the (optional) biometric device. These five levels account for an extremely strong authentication applicable to public networking on public / private computers, and even on TV (satellite, cable, DVD, CD AUDIO, software applications. Transactions including payments may be carried out without any risk of communication tampering, authentication misconduct or identity theft. In essence, the device is a closed box with only two communication ports. The emulation of the device is therefore extremely complex due to the fact that it involves PKI and or identity-based encryption (IBE), key pair, elliptic curves encryption scheme, hardware serialization for communication and software implementation, in conjunction with a specific hardware embodiment and service usage infrastructure component that returns a response necessary for each unique transaction.

The invention discloses a multi-feature cyclic convolution significance target detection method based on an attention mechanism. The method comprises the following steps: ; the method comprises the following steps of: 1, analyzing common characteristics of a salient target in a natural image, including spatial distribution and contrast characteristics, using an improved U-Net full convolutional neural network, performing pixel-by-pixel prediction by adopting an encoder-decoder structure, and performing multi-level and multi-scale characteristic fusion between an encoder and a decoder by adopting a cross-layer connection mode; secondly, a large number of clutters can be introduced to interfere with the generation of a final prediction graph by carrying out concentage fusion on coding end features and decoding end features, so that an attention module is introduced to calibrate full-pixel weights from two angles between channels and between pixels, the task-related pixel weights are enhanced, and the background and noise influence is weakened; and 3, a multi-feature cyclic convolution module is used as a post-processing means, the spatial resolution capability is enhanced through iteration, the edge of an image region is further refined and segmented, and a finer significant target mask is obtained.

A lightweight semantic segmentation method for a high-resolution remote sensing image comprises the steps of network construction, training and testing. Specifically, a deep semantic segmentation network of an encoder-decoder structure is constructed for a pytorch deep learning framework, after network training is carried out based on a remote sensing image data sample set, a to-be-tested remote sensing image serves as network input. A segmentation result of the remote sensing image is obtained. According to the method, on one hand, model parameters are reduced by decomposing depth separable convolution, the calculation complexity is reduced, the semantic segmentation time of the high-resolution remote sensing image is shortened, and the semantic segmentation efficiency of the high-resolution remote sensing image is improved; and on the other hand, semantic segmentation precision is improved through multi-scale feature aggregation, a spatial attention module and gating convolution, sothat the proposed lightweight deep semantic segmentation network can accurately and efficiently realize semantic segmentation of a high-resolution remote sensing image.

A joint code-encoder-decoder design approach and circuit architecture design for (3,k)-regular LDPC coding system implementation. The joint design process relies on a high girth (2,k)-regular LDPC code construction. The decoder realizes partly parallel decoding. The encoding scheme only contains multiplications between sparse matrices and vector and multiplication between a very small dense matrix and vector.

A method, medium, and system compressing and / or reconstructing image data. The data compression method includes compressing current data according to a plurality of modes for compressing current data, determining whether the current data compressed according to the plurality of modes can be represented by a fixed length of bits, selecting a mode from among modes in which the compressed current data can be represented by the fixed length of bits, and outputting the compressed current data according to the selected mode. Accordingly, it is possible to significantly lower the complexity of an image encoder / decoder system, and exactly meet a picture-based Control Bit Rate (CBR) suitable for a Liquid Crystal Display Dynamic Capacitance Compensation (LCD DCC) device / system.

A method and apparatus for encoding / decoding information using low density parity check (LDPC) codewords of varying lengths. In one implementation, an encoder / decoder uses a parity check matrix having a size that corresponds to a length of information for each codeword. In one example, the size of the parity check matrix may be retrieved from multiple stored matrices having different sizes. In other examples, a mother code matrix may be dynamically adjusted by puncturing parity bits and / or deleting information bits. Other inventive embodiments and adaptations are also disclosed.

The invention relates to an omnibearing obstacle detection method based on multi-sensor fusion. The method comprises the following steps: firstly, collecting images and laser point cloud data in different scenes through a laser radar and a camera; carrying out aerial projection on the laser point cloud data, carrying out feature extraction after two-dimensional gridding segmentation, and obtaininga target candidate box in an aerial view; obtaining an image region candidate box by using a one-stage target detection network model; and then fusing the target area candidate box and the image areacandidate box in the aerial view by using spatial registration; designing a segmentation sub-network of an encoder decoder structure in a three-dimensional space point cloud classification branch toclassify each point cloud to obtain an accurate category of an obstacle target in a three-dimensional space; and the three-dimensional candidate box position regression branch calculating the coordinate deviation and loss value of the predicted target and the marked target of the corresponding category, and outputting the deviated predicted obstacle position information to obtain more accurate position information of the three-dimensional space obstacle.