19 results about "Picture reconstruction" patented technology

The present disclosure relates to scalable encoding and decoding of pictures. In particular, a picture is processed by one or more network layers of a trained module to obtain base layer features. Then, enhancement layer features are obtained, e.g. by a trained network processing in sample domain. The base layer features are for use in computer vision processing. The base layer features together with enhancement layer features are for use in picture reconstruction, e.g. for human vision. The base layer features and the enhancement layer features are coded in a respective base layer bitstream and an enhancement layer bitstream. Accordingly, a scalable coding is provided which supports computer vision processing and / or picture reconstruction.

The invention relates to the technical field of graphical user interface generation and rendering, in particular to a multi-terminal-oriented industrial configuration picture self-adaptive presentation system and method, and the system comprises a picture semantic analysis unit, a terminal context acquisition unit, a self-adaptive layout calculation unit and a picture reconstruction and rendering unit. The method comprises the following steps: analyzing an original industrial configuration picture, and generating a picture semantic structure tree containing graphic element attributes and association relationships; collecting target terminal screen parameters, types and interaction modes to form terminal context information; in combination with a preset adaptive rule, calculating an optimal display attribute of the element to generate a layout scheme; and reconstructing the graphic elements according to the scheme and performing layered rendering display. According to the method, multi-terminal precise adaptation of one set of original images is realized, core elements are highlighted, interaction and display effects are optimized, the cost is reduced, the industrial operation efficiency and safety are guaranteed, and the method is suitable for industrial monitoring and control scenes.

The invention provides an attack resisting method, a control device, a storage medium and electronic equipment, and belongs to the technical field of visual language large models. The method comprises the following steps: constructing a picture reconstruction model of a coding-decoding structure, taking a picture encoder of a visual language model as an encoder, and taking a decoder of an MAE architecture as a decoder; training a picture reconstruction model based on the picture data set; inputting an original picture into the trained picture reconstruction model, and generating an adversarial sample based on a projection gradient descent method; and attacking the visual language model by using the confrontation sample pair, and carrying out visual explanation on the attack by comparing the difference between the original picture and the output picture corresponding to the confrontation sample. According to the method, attack countermeasure is carried out based on the visual picture mode and picture reconstruction; and performing visual explanation on the adversarial attack by comparing the change of the reconstructed picture corresponding to the adversarial picture relative to the reconstructed picture corresponding to the original picture.

The application discloses a license plate recognition method, system and computer readable storage medium, and the license plate recognition method comprises the following steps: S1, extracting vehicle picture features to obtain a license plate feature map; S2, generating a license plate position mask according to the license plate feature map; S3, preprocessing the license plate position mask to obtain a first license plate picture; S4, reconstructing the first license plate picture to obtain a second license plate picture; and S5, recognizing the second license plate picture to output a license plate number. According to the application, the license plate feature map is obtained by extracting the vehicle picture features, the license plate position mask is further generated, the first license plate picture is obtained by preprocessing the license plate position mask, and the second license plate picture is obtained by reconstructing the first license plate picture, so that the influence of the angle deviation generally existing in the collected license plate image is reduced, the definition of the license plate picture is improved, and on this basis, the second license plate picture is finally recognized to output the license plate number, so that the accuracy and robustness of the license plate recognition are improved.

The application provides a kind of big data hierarchical encryption transmission optimization method based on edge cloud cooperation, comprising: target identification is carried out to the monitoring video collected, and high-value monitoring area and area label are extracted;Integrating interest area, label, real-time network available bandwidth, edge device computing power load and cloud feedback strategy correction factor, dynamically classify video frame into key event frame, routine activity frame and static background frame;A hierarchical encryption strategy is used to generate an encrypted hierarchical video frame group, which is efficiently uploaded to the cloud according to the transmission strategy generated by the bandwidth and strategy;Complete video is recovered by residual decoding and picture reconstruction;Security feature abstract is extracted from key event frame and is chained and stored;At the same time, the reconstruction completeness is evaluated, and the strategy correction factor is fed back to the edge according to the strategy correction factor, to form a closed loop optimization;For solving the problem of high bandwidth consumption, insufficient security and poor real-time performance in massive data transmission of monitoring video.

A device for decoding video data receives a picture of video data; reconstructs a block of the picture of video data to generate a reconstructed block; and performs a neural network (NN)-based filter process on the reconstructed block to generate a filtered block, wherein the NN-based filter process includes performing a plurality of separable convolutions in parallel with a point-wise input convolution.

Systems, apparatuses, and methods are described for enhancing picture reconstruction in video coding. Bitstream information, such as original bit depth and range indicators, may be used to adjust reconstruction parameters. Compact representations of range differences may be signaled to enable efficient decoding. By adapting clipping ranges for reconstructed samples, the system may improve visual fidelity and reduce artifacts, even when internal processing operates at a different bit depth than the source. These techniques may support flexible configurations and maintain compression efficiency across diverse video formats.

The invention discloses a picture reconstruction method, system and device and a storage medium. The picture reconstruction method comprises the steps of extracting a text feature vector and an image feature vector corresponding to each picture in a picture training data set; fusing the text feature vector and the image feature vector through a first preset fusion method to obtain a first fused feature vector; based on the picture training data set and the first fusion feature vector, training a preset first initial picture reconstruction model to obtain a first picture reconstruction model; performing element-by-element multiplication on the text feature vector and the image feature vector to obtain a second fusion feature vector; and training the first picture reconstruction model based on the picture training data set and the second fusion feature vector to obtain a second picture reconstruction model, thereby improving the picture reconstruction accuracy of the model.

Reconstruction of a frame or picture using adaptive resolution is described. A coded unit of a picture coded at a reduced resolution is reconstructed. The picture including the reconstructed coded unit is stored in a picture buffer at the reduced resolution. The picture is resampled to full resolution and a filter with block level signaled parameters is applied to the picture at full resolution.

The invention particularly relates to a nuclear power plant DCS two-layer system operator station function mirror image implementation method and system, and belongs to the technical field of control systems. The method comprises the following steps: analyzing a static picture file and configuration information of a nuclear power plant DCS two-layer system; offline reconstruction of a human-computer interface system is realized; establishing data communication with a PI database, and realizing dynamic display of image data of a human-computer interface system; on the basis of a nuclear power plant DCS two-layer system picture, an equipment basic database, a DCS configuration database, an EAM completion report database and an experience feedback database are associated, and multi-position integrated digital information display is achieved. The system realizes the method. According to the invention, the picture reconstruction of the DCS two-layer system of the nuclear power plant and the remote display of the multi-in-one DCS digital information are realized, the number of terminals is not limited, and the operation of accessing the server by multiple users at the same time is realized.

The application provides a fingerprint retrieval method and device, the method comprising: constructing a contrast learning network; reading a plurality of first pictures, and performing picture reconstruction on each first picture to form a second picture respectively, wherein the first pictures contain first fingerprint information; inputting the plurality of first pictures and the second pictures into the contrast learning network to obtain first output data and second output data corresponding to each pair of first picture and second picture respectively; calculating the similarity of each corresponding first output data and second output data; updating the weight information of a first network in the contrast learning network based on the similarity; updating the weight information of a second network in the contrast learning network based on the updated weight information of the first network, thereby obtaining a fingerprint identification network storing first fingerprint information on all the first pictures; inputting second fingerprint information into the fingerprint identification network; and completing the retrieval of the second fingerprint information based on at least the fingerprint identification network.

A method performed by a decoder for generating a reconstructed picture corresponding to an original picture, wherein the original picture consists of a plurality of regions including at least a first region and a second region. The method includes obtaining a bitstream. The method also includes producing the reconstructed picture using the bitstream. The bitstream includes a first coded picture sample unit corresponding to the first region of the picture, and the bitstream further includes indicator information indicating that, for the second region of the picture, the bitstream does not include a coded picture sample unit corresponding to the second region of the picture.

The present disclosure relates to scalable encoding and decoding of pictures. In particular, a picture is processed by one or more network layers of a trained module to obtain base layer features. Then, enhancement layer features are obtained, e.g. by a trained network processing in sample domain. The base layer features are for use in computer vision processing. The base layer features together with enhancement layer features are for use in picture reconstruction, e.g. for human vision. The base layer features and the enhancement layer features are coded in a respective base layer bitstream and an enhancement layer bitstream. Accordingly, a scalable coding is provided which supports computer vision processing and / or picture reconstruction.

The invention provides a digital semantic communication method and system for image transmission, and the method comprises the steps: S1, inputting an original image into a semantic coding neural network, and extracting the semantic features of the original image; s2, inputting the semantic features into a probability generation neural network, and generating a probability table corresponding to the semantic features; s3, the probability table is input into a Gumbel-Softmax sampling layer, and a discrete semantic bit sequence is generated; s4, transmitting the semantic bit sequence through a bit channel; and step S5, inputting the received semantic bit sequence into a semantic decoding neural network for picture reconstruction, and finally reconstructing a picture. According to the method, discrete bits and constellation points are decoupled, a Gumbel-Softmax sampling method is used, a quantization module is omitted, an optimal mapping relation from information source data to a transmission bit sequence is directly learned by using a neural network, and a semantic communication system is enabled to be better compatible with an existing communication system on the premise of not losing performance.

The invention discloses a Fresnel aperture coding imaging method and device based on two-stage reconstruction and noise matching, and the method comprises the steps: carrying out the convolution operation of a preprocessed real image and a Fresnel aperture, and obtaining a simulation image; brightness conversion and noise fitting operation are carried out on the simulation image, so that noise distribution of the simulation image is close to real data, and a coding graph is obtained; performing back propagation reconstruction on the coded image to obtain an initial reconstructed image; according to the fast Fourier transform module and the convolutional neural network, constructing a target FFTConv-UNet model; and inputting the initial reconstructed image into the target FFTConv-UNet model to obtain a target reconstructed image. The method can improve the quality of picture reconstruction, and can be widely applied to the technical field of image reconstruction.

In a method of video decoding, a coded video bitstream including coded information of a plurality of coded pictures is received. The coded information includes a supplemental enhancement information (SEI) message associated with a first picture and a second picture in the plurality of coded pictures, and the SEI message includes a syntax element indicative of a truncation function that converts the first picture of a first bit depth to a second picture of a second bit depth, the second bit depth being lower than the first bit depth. In the method, a reconstructed picture corresponding to the second picture is generated. The reconstructed picture includes corresponding reconstructed samples based on the coded information. A range of sample values of the reconstructed samples of the reconstructed picture is determined by the truncation function.

Systems, apparatuses, and methods are described for enhancing picture reconstruction in video coding. Bitstream information, such as original bit depth and range indicators, may be used to adjust reconstruction parameters. Compact representations of range differences may be signaled to enable efficient decoding. By adapting clipping ranges for reconstructed samples, the system may improve visual fidelity and reduce artifacts, even when internal processing operates at a different bit depth than the source. These techniques may support flexible configurations and maintain compression efficiency across diverse video formats.

A method for decoding a bitstream includes: receiving a bitstream and decoding, using coded information of the bitstream, one or more pictures. The decoding of the one or more pictures includes: determining whether a generative face video supplemental enhancement information (SEI) message matches with a generative network; and in response to the generative face video SEI message matches with the generative network, decoding the SEI message. The decoding of the SEI message includes: determining a face information parameter and a base picture associated with the SEI message; and reconstructing a face picture based on the face information parameter and the base picture.

A method comprising: encoding an input picture into a coded picture; reconstructing a decoded picture corresponding to the coded picture; encoding a spatial region into a coded tile, the encoding comprising: determining a horizontal offset and a vertical offset indicative of a region-wise anchor position of the spatial region within the decoded picture; encoding the horizontal offset and the vertical offset; determining that a prediction unit at position of a first horizontal coordinate and a first vertical coordinate is predicted relative to the region-wise anchor position; indicating that the prediction unit is predicted relative to a prediction-unit anchor position; deriving a prediction-unit anchor position equal to sum of the first horizontal coordinate and the horizontal offset, and the first vertical coordinate and the vertical offset, respectively; and determining a motion vector for the prediction unit; and applying the motion vector relative to the prediction-unit anchor position to obtain a prediction block.