47results about How to "Improve decoding quality" patented technology

An automatic dual-step, text independent, language-independent speaker voice-print creation and speaker recognition method, wherein a neural network-based technique is used in a first step and a Markov model-based technique is used in a second step. In particular, the first step uses a neural network-based technique for decoding the content of what is uttered by the speaker in terms of language independent acoustic-phonetic classes, wherein the second step uses the sequence of language-independent acoustic-phonetic classes from the first step and employs a Markov model-based technique for creating the speaker voice-print and for recognizing the speaker. The combination of the two steps enables improvement in the accuracy and efficiency of the speaker voice-print creation and of the speaker recognition, without setting any constraints on the lexical content of the speaker utterance and on the language thereof.

The invention belongs to the field of three-dimensional (3D) video coding technologies and provides a novel multi-view video fractal coding, compressing and decompressing method. According to the method, mass data of a multi-view 3D video is efficiently compressed so as to be stored and transmitted; in a coding process, the time and space combined prediction is realized, and a video compressing method in which quick disparity estimation and center-biased pentagonal motion estimation/prediction are combined with fractal coding is provided, so that the data redundancy is effectively reduced; a five-view video is taken as an example, K, L, C, R and S sequentially represent a video view point respectively, a prediction structure of C to L to K and C to R to S is provided, namely that both the view point R and the view point L are predicted by using the view point C and the view point S and the view point K are respectively predicted by using the view point R and the view point L, and start frames of the five view points are all frames I; and view point decoding sequences are same, and the blocking effect is removed by adopting a loop filter. The method has the advantages that under the conditions that better video decoding quality is obtained, the coding rate and compression ratio are greatly increased, the operational complexity is reduced, and a foundation for real-time application of multi-view video coding is laid.

The invention discloses a video transmission error control method and a video transmission error control system. The method comprises the error code masking process which comprises the steps of: performing edge intensity quantitative analysis on damaged macroblocks so as to divide the damaged macroblocks into smooth blocks or edge blocks; for the smooth blocks, performing error code masking by adopting a bilinear difference algorithm; and for the edge blocks, performing error code masking by adopting a directional interpolation algorithm. The method and the system optimize video transmission image decoding quality and improve video decoding effect.

The invention provides a video compression and decompression method based on fractal and H.264. Intra prediction of coding based on the H.264 is adopted for an I frame to obtain a prediction frame. A block motion estimation or complementary coding which are based on fractal is used for obtaining a prediction frame of a P frame. Difference between each initial frame and each corresponding prediction frame is a residual frame. After the residual frames are transformed through discrete cosine transformation (DCT) and are quantified, on one hand, the residual frames are written in a code rate, on the other hand, the residual frames are inversely quantified and reversely transformed though the DCT and then plus the prediction frames to get rebuilt frames (as reference frames). Fractal parameter is generated when the P frame is encoded in a predictive mode, the residual frames of the I frame and the P frame are compressed by the utilization entropy encoding CALCV, and the fractal parameter is compressed by utilization of an exponential-Golomb code with symbols, and a corresponding decompression process is that a prediction frame is obtained from an intra prediction of the I frame, and a prediction frame is obtained by a prediction of a frame in front of the P frame. Residual information in the code rate is reversely quantified and reversely transformed through the DCT to respectively obtain the residual frames of the I frame and the P frame.

The invention provides a multi-view stereoscopic video compression and decompression method based on fractal and H.264. A middle view is used as a base layer and encoded on the principle of a motion compensated prediction (MCP) method in a compression mode, and other views are encoded on the principle of MCP + digitally controlled potentiometers (DCP) in a compression mode. Three views are taken for example, intra prediction coding of the H.264 is adopted by an I frame of the middle view, a rebuilt frame (as a reference frame) is obtained after filtration, a P frame of the middle view is encoded with block MCP by the utilization of the encoded rebuilt frame as the reference frame, the most matched fractal parameter of the block of each time is recorded, the fractal parameters are respectively substituted in a compression iterative function system to achieve a prediction frame of the P frame, and the rebuilt frame is obtained after filtration. The MCP + DCP is adopted for coding by a left view and a right view and a reference frame can be an encoded front frame of the respective view or an encoded frame of other views. Entropy encoding CALCV is used for compressing a residual frame of each frame and an exponential-Golomb code with symbols is used for compressing the fractal parameter.

The invention discloses a method for decoding dynamic Viterbi, comprising the following steps: using set parameter values to decode an existing receiving sequence based on a Viterbi algorithm to acquire decoding, wherein the parameter values comprise memory depth and/memory width, and the memory depth represents backtracking span of arrow diagrams in the Viterbi algorithm and/or the memory width represents branch metric of the arrow diagrams in the Viterbi algorithm; adjusting the parameter values in the range of adjustment lower limit and adjustment upper limit; and using the Viterbi algorithm which is used for adjusting the parameter values to decode the next receiving sequence. The invention also discloses a device for decoding dynamic Viterbi. According to the invention, decoding parameters can be adjusted dynamically in accordance with the condition of existing interchannel noise, the decoding performance of the Viterbi can be exerted maximally, the error rate is reduced effectively, and the effective utilization of the resources can be ensured at the same time.

An auto-exposure (AE) control system, includes a camera configured to capture an input image, a region of interest (ROI) determination circuit configured to determine a ROI of the input image, a depth decoder and an AE controller. The depth decoder is configured to generate a depth map of the input image and determine a decode rate value according to the depth map of the input image and the ROI of the input image. The AE controller includes an exposure adjustment circuit which is configured to receive a decode rate value and the detection result and sequentially adjust a plurality of the exposure parameters according to the variation of the decode rate value. A step size for adjusting the plurality of the exposure parameters is determined according to the variation of the decode rate value and the detection result.

A method of parametric coding of a multichannel digital audio signal including coding a signal arising from a channels reduction processing applied to the multichannel signal and coding spatialization information of the multichannel signal. The method includes the following acts: extraction of a plurality of items of spatialization information of the multichannel signal; obtaining at least one representation model of the extracted spatialization information; determination of at least one angle parameter of a model obtained; coding the at least one determined angle parameter so as to code the spatialization information extracted during the coding of spatialization information. Also provided are a method for decoding such a coded signal and corresponding coding and decoding devices.

A variable length decoding apparatus and method for the image format of a digital video camera (DV) is provided for decoding the images of a digital video camera on the PASS2 decoding layer. The apparatus comprises a first register, a second register, a third register and a fourth register for saving a bit stream, a barrel shifter for producing output data with a length of sixteen bits, a bit stream generator for generating the combined bit stream, a programmable logic array for decoding the image by looking up the table, a controller for controlling the decoding process and fixed length decode, and an accumulator for accumulating the bit number of the bit stream that is being decoded. The present apparatus can decode much higher frequency components, and the quality of the decoded image is better and the color of the decoded image is more vivid.

The invention discloses a training method and a decoding method for a decoder of a generative dialogue system. The method comprises the following steps: 1) for each question code in a question code set, predicting the question code by using forward and backward neural networks to respectively obtain vector representation results; 2) calculating the difference between prediction results of each step of the forward and backward neural networks, and taking the difference as a loss function of the encoder of the generative dialogue system; 3) calculating the state difference of each step of the forward and backward neural networks to serve as the local difference of the forward and backward neural networks; 4) calculating sentence vector representations of vector representation results generated by the forward and backward neural networks, and calculating a difference between the two sentence vector representations as an integral difference between the two sentence vector representations;5) adding the local difference and the integral difference as penalty term functions into a loss function to obtain an integral penalty function as a loss function of the encoder of the generative dialogue system, and 6) predicting the question codes by using the trained forward neural network to generate reply contents.

A device relating to information processing technologies and including an encoding and decoding method configured to solve the poor decoding quality problem. The method includes: encoding each sample of an input signal to generate an encoded signal of a core layer; comparing residuals of all or a part of the samples of the input signal with encoding thresholds, where the residuals are generated by core layer encoding, and performing encoding according to comparison results to generate an encoded signal of an enhancement layer; and writing the encoded signal of the core layer and the encoded signal of the enhancement layer into a bitstream to generate an encoded signal of the input signal.

This invention relates to a mistake processing method and device, used for decoding data, wherein the data comprises an image group, and the image group comprises several associated images. The method comprises the following steps: a) checking if the received image group is continuous; b) if it is not continuous, making sure the type of the lost image; c) processing the image group according to the type and the corresponding correct rule. The type of the image comprises the image which needs to be consulted or not to be consulted when decoding other image. By said invention, it can decode even the image head information is lost.

The present invention discloses a method for coding an image, a method for decoding an image, a method for processing image data, and devices thereof. The method for coding the image includes: determining a current intra coding block that uses a constrained intra prediction method among current inter coded frames; determining that neighboring coding blocks that are adjacent to the current intra coding block include a neighboring inter coding block; updating image data in the neighboring inter coding block based on an invariable value or image data in a neighboring intra coding block included in the neighboring coding blocks; and coding the current intra coding block based on the updated image data in the neighboring inter coding block. By using the methods and devices, the error recovery capability of an intra coding block may be improved, and the image data decoding quality may be improved.

The invention discloses an H.264/AVC (Any Video Converter) video scheduling method based on double decoding simulation. The method comprises the following steps of (1) calculating the packet quality contribution index (PQCI) of each H.264/AVC video packet to be transmitted, wherein PQCI is obtained by calculating two kinds of distortion, and each kind of distortion comprises two parts: the quality contribution of the video packet to a current frame and the quality contribution of the video packet to a subsequent frame; (2) during the calculation of the own quality contribution of each video packet, bringing two decoders into a transmitting end for performing distortion estimation by adopting a method based on double decoding simulation, and solving quality contribution; (3) during the calculation of the quality contribution of each video packet to the subsequent frame, solving by adopting a diffusion factor weighting method; and (4) sequencing and scheduling video packets to be transmitted according to the magnitudes of PQCIs, and transmitting the video packets. Due to the adoption of the H.264/AVC video scheduling method, the quality of videos decoded at a receiving end is improved, the performance of video transmission in a wireless network is improved, and user experience of the receiving end is improved finally.

This invention discloses a dual purpose digital audio-frequency broadcast receiver/digital music broadcast device. It is composed of: in proper order connected a tuning, an ADC connecting the output end of CPU, a digital broadcast decoder (DBCD) connecting the output end of the CPU, a DAC connecting the output end of the DBCD, a loudspeaker connecting the output end of the DAC. It comprises also: a digital music memory and a digital music decoder connecting respectively with the CPU, an ear-phone connecting the output end of the DAC, DAC connecting CPU, the output end of the digital music decoder connecting the input end of the DAC. Advantages are: new application of broadcast receiver, being used for also receiving music with lowered product cost.