75 results about "Resolution recovery" patented technology

An image not subjected to color correction, gamma correction, MTF correction and others is used as an image to be subjected to Retinex image processing. Thus, a RAW image prepared only by converting an electric signal of a CCD image sensor by an A / D converter into a digital form is used as an image to be subjected to the Retinex processing. Predetermined gamma correction is performed after the Retinex processing. Thereby, Retinex algorithm substantially modeling functions of a retina and a cortex of a living body exhibits its original performance. The resolution of the original image is lowered, and processing is performed on an image thus produced to produce a blurred image. Then, the resolution of the blurred image is restored to the same resolution as that of the original image, and the Retinex processing is performed. Thereby, the Retinex processing can be performed without increasing memory consumption and extremely lowering an operation speed.

The invention discloses an intelligent traffic analysis system, comprising a parameter setting module for setting the control parameters and calibration parameters required by each module of the system; a resolution conversion module for converting the control parameters, the calibration parameters and the resolution of an input collected image; a parameter represent module for representing the control parameters and the calibration parameters after resolution conversion into the parameter types applicable in other processing modules; a processing module for receiving the represented parameters and an object image, to process the object image under the control of the represented parameters to obtain processing result; a resolution recovery module for converting the processing result to the resolution of the input image to obtain output result; and an output module for outputting the output result. The intelligent traffic analysis system has complete structure and simple process, and is suitable for various processing algorithms, thus obtaining strong expansibility.

A system and method for recovering resolution in an x-ray computed tomography system. The system includes an x-ray source, a two-dimensional sensing array, and a controller. The x-ray source projects an x-ray beam towards subject. The x-ray source includes an anode and a cathode, where the anode has a target for directing the x-ray beam. A two-dimensional sensing array receives an x-ray beam through the subject to generate an image. The target of the anode causes a shift variant blur of the x-ray beam received by the sensing array due to the angle between the target and the sensing array. The controller is in communication with the sensing array to control acquisition and processing of the image. The controller is configured to adjust for the shift variant blur of the x-ray beam.

The invention discloses a high-quality depth estimation method based on depth prediction and an enhanced sub-network, and belongs to the field of image processing and computer vision. The method comprises performing depth prediction on a color map by constructing a depth prediction sub-network, and using a depth enhanced sub-network to recover the resolution of a low-resolution image obtained by the depth prediction sub-network, thereby obtaining a high-resolution depth prediction map. The method is simple and easy to implement. The system is easy to construct, and can obtain a corresponding high-quality depth map from a single color map end to end by using a convolutional neural network. The information lost by the downsampling of the depth prediction sub-network is restored by the high-frequency information of the color image, and a high-quality and high-resolution depth map is finally obtained. The spatial pooling pyramid structure effectively solves the problem of predictive accuracy caused by different object sizes.

The invention discloses a cGAN algorithm-based super-resolution image recovery technology, and relates to super-resolution recovery for deep convolutional antagonistic neural network images. The existing method needs to carry out a lot of training, the network layers are relatively shallow and the feature weight values included in models obtained through the training are not complete enough, so that the final recovery effect is not ideal. Greyscale maps are used during the training, so that the super-resolution recovery effect for three-channel colored images is not good. Aiming at the defects in the prior art, the invention provides a cGAN algorithm-based super-resolution image recovery method which is capable of training the colored images to achieve better recovery effect. The key points of the method comprises that (1) the training time is short, (2) three-channel colored maps can be directly trained, (3) much image preprocessing is not required, and (4) the models obtained through training can be used once for all.

The invention discloses a human face super-resolution method based on residual keeping. According to the human face super-resolution method, based on a learning method, high and low resolution training image pairs are partitioned uniformly first, and processing is conducted in a block set based on positions: the smooth high frequency component and the standard deviation high frequency component are estimated on the corresponding positions of the training block set for all blocks of an input image, and the high-frequency compensation component is estimated through a residual training set. The linear combination of the three high-frequency information is used for reconstructing a high-resolution image on a high-resolution average face, and therefore the noise problem in super-resolution restoration of a human face image (like a monitoring image) with serious noisy points can be solved.

An image not subjected to color correction, gamma correction, MTF correction and others is used as an image to be subjected to Retinex image processing. Thus, a RAW image prepared only by converting an electric signal of a CCD image sensor by an A / D converter into a digital form is used as an image to be subjected to the Retinex processing. Predetermined gamma correction is performed after the Retinex processing. Thereby, Retinex algorithm substantially modeling functions of a retina and a cortex of a living body exhibits its original performance. The resolution of the original image is lowered, and processing is performed on an image thus produced to produce a blurred image. Then, the resolution of the blurred image is restored to the same resolution as that of the original image, and the Retinex processing is performed. Thereby, the Retinex processing can be performed without increasing memory consumption and extremely lowering an operation speed.

The embodiment of the invention discloses a rendering method and apparatus, an engine and a storage medium. The rendering method includes the steps: acquiring a low resolution rendering graph and a target resolution texture graph of the target scene; based on a preset super-resolution recovery algorithm, performing super-resolution recovery on the low resolution rendering graph so as to generate an initial target resolution rendering graph; and based on a preset filtering algorithm, according to the target resolution texture graph, performing filtering on the initial target resolution rendering graph so as to generate a result rending graph. The rendering method solves the problem that the prior art consumes longer time for generating the result rendering graph of the target resolution, sothat the result rendering graph carries high resolution texture and also carries the light sensation information corresponding to overall arrangement of the texture of the target resolution. Besides,compared with the prior art which generates a high resolution rendering graph only through rendering, the rendering method greatly reduces the rendering time.

The invention provides an adaptive training library-based block domain face super-resolution reconstruction method. The new method based on learning comprises the following steps of: firstly, uniformly dividing high resolution images and low resolution training images into blocks, and clustering in a block set for which position is considered as the unit; and then, adaptively selecting corresponding position training block set for each block of the input images, screening the adaptive training set according to neighboring selection and adaptive class selection, and recovering high frequency details in a low resolution block by executing principal component analysis coefficient projection in the adaptive training set. By the method, the problem of noise in super-resolution recovery of the face image (such as a monitoring image) with serious noise points is solved or relieved.

The invention provides a robustness face super-resolution processing method based on contour inspection, which comprises the following steps: on the basis of a traditional face super-resolution method based on manifold, simultaneously selecting two types of image features for mass reduction process robustness and simultaneously taking as a criterion of image similarity in the algorithm process through coupling the two types of image features into an (Locally embedding) algorithm. The robustness face super-resolution processing method solves the problem incapable of being better solved by a traditional method that serious noise and fuzziness are lack of reality to a super-resolution recovery effect of a single frame face image in the monitoring and imaging process.

A method and apparatus for compensating for motion in an imaged object are provided. The method includes obtaining nuclear medicine (NM) image data for an object acquired from an NM imaging system, wherein the NM image data includes single photon emission computed tomography (SPECT) image data. The method further includes performing resolution recovery on the NM image data. The method also includes reconstructing the resolution recovered NM image data into a plurality of subsets and morphing the reconstructed NM image data in at least one of the plurality of subsets. The method additionally includes normalizing the NM image data based on a magnitude of the morphing and summing the NM image data to generate an NM image compensated for motion of the object.

The invention discloses a method for learning super-resolution network based on sub-band of group structure in wavelet domain, which realizes super-resolution restoration of single picture by constructing a novel super-resolution network comprising a feature extraction network and an image restoration network. The input low-resolution image is represented as a set of feature maps by a feature extraction network. The image restoration network includes a cluster upsampling module and a convolution layer. The output features of the feature extraction network are transferred to the cluster upsampling module for sub-band extraction, self-residual learning and sub-band refinement. The inverse discrete wavelet transform is used to change the resolution of the feature image so as to recover the feature image with high resolution. Then the high resolution image is restored by convolution layer. The method of the invention has good quantization index and can effectively realize super-resolutionrestoration of a single picture.

The invention relates to a tumor imaging quantitative detection technology, in particular to a method for quantitatively measuring a tumor standard uptake value, a thin body standard uptake value, a metabolic volume and total glucolysis through Coincidence SPECT or SPECT / CT and application of the method in tumor assessment. The implementation steps of the method comprise an image acquisition step, a scatter correction step, a nuclide physical decay correction step, a movement correction step, a tissue attenuation correction step, an image spatial resolution recovery step, a noise removal step, a standard uptake value calculation step and a tumor therapeutic effect assessment step. By means of the technological means of the Coincidence SPECT tumor imaging quantitative analysis technology and the application in tumor assessment, the clinical problem of quantitatively measuring the tumor standard uptake value, the thin body standard uptake value, the metabolic volume and the total glucolysis is solved through the Coincidence SPECT and the SPECT / CT, and the technology can be used in the clinical assessment of tumor.

The invention discloses a video coding method based on deconstruction compression and fusion. According to the method, based on redundancy of resolution information in video coding, a video sequence frame is deconstructed into a high-resolution spatial domain texture frame and a low-resolution time domain motion frame for coding, and the decoded low-resolution time domain motion frame is recoveredto a high-fidelity original resolution by using a super-resolution method based on deep learning, thereby improving the encoding efficiency enhancement. The method comprises the following specific steps: (1) generating an image block combination used for training as a supervised training data set; (2) constructing a loss function, and independently training the super-resolution networks under different compression ratios by using an optimizer; (3) constructing a loss function, and optimizing the whole network end to end by using an optimizer; (4) using the optimized super-resolution network to perform super-resolution on the low-resolution time domain motion frame in the deconstructed and encoded video, recovering the image texture and motion details under the original resolution, and improving the overall video encoding efficiency.

In one embodiment, a resolution resiliency application performs robust domain name system (DNS) resolution. In operation, the resolution resiliency application determines that an authoritative name server that is responsible for a domain name specified in a DNS query is unavailable. In response to determining that the authoritative name server is unavailable, the resolution resiliency application performs operation(s) that modify one or more DNS records stored in a cache based on one or more resiliency policies associated with the authoritative name server. The resolution resiliency application then generates a DNS response to the DNS query based on a DNS record stored in the modified cache. Notably, unlike conventional techniques that may generate inaccurate DNS responses based on stale DNS records, the disclosed techniques increase the likelihood of providing clients with DNS responses that accurately provide requested information.

The embodiment of the invention discloses an obtaining method and device of a rendering image, a rendering system and a storage medium. The method comprises the steps of obtaining a low-resolution rendering image and a target-resolution texture image which are generated by a rendering engine of a target scene; generating a low-resolution texture image according to a preset rule; extracting light sensation distribution change information of the low-resolution rendering image relative to the low-resolution texture image to generate a light sensation distribution change image; conducting super-resolution restoration on the light sensation distribution change image to generate a target-resolution light sensation distribution change image; merging the target-resolution light sensation distribution change image with the target-resolution texture image to generate a resultant rendering image. The problem in the prior art that generation of the resultant rendering image with a high resolutionis time-consuming is solved, the technical effects that the rendering effect is guaranteed, and the rendering time is shortened at the same time are achieved, and the rendering efficiency is improvedthrough the cooperation of the rendering engine.

A nuclear camera (10) includes a plurality of detector heads (12) which have collimators (14) for fixing the trajectory along which radiation is receivable. A rotating gantry (22) rotates the detector heads around the subject collecting less than 360° of data, e.g., 204° of data. A zero-filling processor (50) generates zero-filled projection views such that the actually collected projection views and the zero-filled projection views span 360°. A smoothing processor (56) smooths an interface between the zero-filled and actually collected projection views. The zero-filled and smoothed views are Fourier transformed (60) into frequency space, filtered with a stationary deconvolution function (62), and Fourier transformed (64) back into real space. The resolution recovered projection data sets in real space are reconstructed by a reconstruction processor (68) into a three-dimensional image representation for storage in an image memory (70).

The current invention presents designs of SPECT gamma cameras without the in-out mechanical motion of the detectors. The elimination of this motion is achieved by the implementation of iterative algorithms, such as Resolution Recovery and / or Wide Beam Reconstruction, which compensate for the Line Spread Function effect due to the collimator characteristics. The use of these methods enables construction of SPECT gamma cameras with a range of novel designs, having their gamma detector (or detectors) orbiting the patient in a predetermined orbit of fixed radius. For example, the radius might be chosen as such that the majority of all patients can be scanned by the system. The advantages of the invention are applicable for gamma cameras with any numbers of detectors.

The invention relates to a video super-resolution recovery method based on deep learning and adjacent frames. The method comprises the following steps: (1) data preprocessing: (2) data enhancement: (3) data conversion: (4) network architecture construction: a network comprises a frame alignment module, a frame fusion module and a reconstruction module; (5) inputting each current video image frameneeding to be subjected to super-resolution recovery and n frames (2n + 1 frames in total) of images before and after the current video image frame into the network architecture constructed in the step (4); and outputting the recovered super-resolution video. According to the invention, the obtained image has a better effect.

The embodiment of the invention discloses a rendering method, device and engine, and a storage medium. The rendering method comprises: obtaining a low-resolution rendering graph, a target resolution texture graph and a low-resolution texture graph of a target scene, wherein the resolutions of the low-resolution texture graph and the low-resolution rendering graph are the same; extracting the light sensation distribution changing information of the low-resolution rendering graph relative to the low-resolution texture graph to generate a light sensation distribution change chart; performing super-resolution recovery of the light sensation distribution change chart, and generating a target resolution light sensation distribution change chart, wherein the resolutions of the target resolution light sensation distribution change chart and the target resolution texture graph are the same; and fusing the target resolution light sensation distribution change chart and the target resolution texture graph, and obtaining a result rendering graph. The problem is solved that the generation of the result rendering graph with high resolution is time-consuming in the prior art, and the technical effect is achieved that the rendering time is reduced while the rendering effect is ensured.

In one embodiment, a resolution resiliency application performs robust domain name system (DNS) resolution. In operation, the resolution resiliency application determines that an authoritative name server that is responsible for a domain name specified in a DNS query is unavailable. In response to determining that the authoritative name server is unavailable, the resolution resiliency application performs operation(s) that modify one or more DNS records stored in a cache based on one or more resiliency policies associated with the authoritative name server. The resolution resiliency application then generates a DNS response to the DNS query based on a DNS record stored in the modified cache. Notably, unlike conventional techniques that may generate inaccurate DNS responses based on stale DNS records, the disclosed techniques increase the likelihood of providing clients with DNS responses that accurately provide requested information.

The invention relates to a DSM local missing repairing method based on deep learning, and belongs to the technical field of DSM data processing. According to the method, a repair model comprising a feature extraction module, a channel fusion module and a resolution recovery module is constructed, and the feature extraction module performs feature extraction by adopting partial convolution, so thatthe shape of a convolution kernel can be randomly changed according to the shape of a mask, and the extraction capability of irregular missing edge features is enhanced; the channel fusion module adds feature weights in two dimensions of a channel and a space, and important features are selected for restoration; and the resolution recovery module recovers the resolution of the feature map in an up-sampling and partial convolution mode. By means of the restoration model, the restoration precision is higher on the whole, and meanwhile better robustness is achieved in the aspect of missing proportion change.

The invention relates to a remote sensing image building extraction method based on an attention mechanism and boundary loss. The method comprises the following steps: acquiring a to-be-extracted optical remote sensing image; calling the trained building extraction model; and inputting the optical remote sensing image into the building extraction model for building extraction to obtain a building extraction result of the optical remote sensing image. The encoder of the building extraction model comprises a feature extraction layer used for carrying out feature information extraction on an optical remote sensing image; the global context attention module is used for performing global context feature fusion processing on the feature information output by the feature extraction layer; and the cavity space pyramid pooling module is used for carrying out cross-scale fusion on the feature maps output by the feature extraction layer. A decoder of the building extraction model is used for carrying out resolution recovery on a feature map output by an encoder, and a loss function is a combined loss function of a boundary loss function and a Dice loss function. And the separability of the extracted building is greatly improved.

An image transmitting apparatus includes a region detecting unit detecting, in an image, a region of interest matching a predetermined condition; a resolution converting unit selecting, for reducing resolution of the image, an image passed through a low-pass filter as an image outside the detected region of interest and an image not passed through the low-pass filter as an image inside the region of interest; and a transmitting unit transmitting the image with reduced resolution and positional information of the region of interest. An image receiving apparatus includes a resolution restoring unit performing image quality improvement processing on the region of interest obtained from the positional information and included in the image transmitted from the image transmitting apparatus so as to increase resolution in the region of interest; and a combining unit combining the image inside the processed region of interest with the image outside the region of interest.

The invention discloses an image reversion method with the image errors and small bore diffraction image quality regressive imaging system. It include the processes: (1) after scene imaging through the imaging lens of small bore with image error, the image is degenerated by CCD and output the degeneration image to image processing unit; (2)the unit recovers the image to the high resolution factor image according to the lens imaging character. The image reversion method of invention uses the optical system information (optical caliber, image error and all kinds of machining error), so the degeneration module is precision and the user can get the good super resolution factor recovery method but not knowing the camera parameter, so the recovery effect is better than the normal method.

The invention relates to a method and system for conducting image measurement to obtain the tumor standardized uptake value through nuclear medicine single photons. Image measurement is conducted to obtain the tumor standardized uptake value through single-photon emission computed tomography (SPECT) or obtain the tumor standardized uptake value through single-photon emission computer tomography / computer tomography (SPECT / CT). The method includes the photography step, the scattering correction step, the attenuation correction step, the image resolution ratio recovery step, the noise removal step and the uptake value calculation step. The method solves the problems of physical interference on dynamic single-photon emission computer tomography and patient movement, and therefore the capacity for quantifying the tumor standardized uptake value of patients can be achieved.

A resolution adjustable game controller has a resolution button and a linear controller. When the resolution button is pushed, the resolution of the linear controller is changed to another resolution, and when the resolution button is released, the resolution restores. The resolution button is one of the buttons originally on the game controller or an additional button different from the original buttons on the game controller. The control circuit is a chip which is electrically connected to the resolution button for controlling the resolution. When the resolution button is pushed, the resolution button is triggered, when the resolution button is released, the trigger condition is also released.

The invention discloses a method for resolving a barnidipine mother nucleus by using glucosamine as a resolving agent. The method comprises the following process routes shown in the description: resolving the barnidipine mother nucleus by using the glucosamine as the resolving agent, wherein the easily available low-cost barnidipine is used as a chiral resolving agent for resolving the racemized barnidipine mother nucleus and solvents used in the middle are easily available and low-cost alcohol and water; and obtaining a single isomer of the barnidipine mother nucleus with high yield and high optical purity (ee value is 99.0-99.6%) with. The method has the beneficial effects that the resolving agent is low-cost and easily available, the using of expensive resolving agents is avoided, the used solvents are harmfulless, the resolution recovery rate is high, and the product has high optical purity and strong atom economy; the experiment operation is simple, reaction conditions are mild and easy to control, and the technology repeatability is high; and the implementation of the industrial production is easy.

The embodiment of the invention provides an image data transmission method, system and device, electronic equipment and a storage medium, and relates to the technical field of information processing.The method comprises carrying out down-sampling on a to-be-processed image at a server; inputting the to-be-processed image into a pre-trained super-resolution model for super-resolution recovery; calculating residual error information according to the image after super-resolution recovery and the to-be-processed image; sending the down-sampled image and the residual error information to a clientso as to save the bandwidth, and inputting the down-sampled image into a pre-trained super-resolution model which is the same as the server at the client to carry out super-resolution recovery on thedown-sampled image and then obtain a super-resolution restored image; and compensating the super-resolution restored image according to the residual information to obtain a final image, and performingimage compensation through the residual information to obtain the final image, so that the bandwidth can be saved under the condition of reducing image distortion as much as possible.