In one embodiment, an apparatus comprises a storage device and a processor. The storage device stores a plurality of images captured by a camera. The processor: accesses visual data associated with an image captured by the camera; determines a tile size parameter for partitioning the visual data into a plurality of tiles; partitions the visual data into the plurality of tiles based on the tile size parameter, wherein the plurality of tiles corresponds to a plurality of regions within the image; compresses the plurality of tiles into a plurality of compressed tiles, wherein each tile is compressed independently; generates a tile-based representation of the image, wherein the tile-based representation comprises an array of the plurality of compressed tiles; and stores the tile-based representation of the image on the storage device.

The present disclosure discloses a method, device, electronic equipment and storage medium for generating a super night scene image. The method includes the following steps: acquiring consecutive multiple frames of original images, which include a frame of underexposed image and multiple frames of normally exposed images; performing stacked noise reduction processing on the multiple frames of normally exposed images to obtain a frame of normally noise-reduced image; performing gray scale transformation processing on the normally noise-reduced image to obtain a frame of overexposed image; fusing the underexposed image, the normally noise-reduced image and the overexposed image to obtain a frame of super night scene image.

A video denoising method includes: while continuing to receive a video stream, performing multi-stage denoising processing on a respective frame, including: detecting a change in a current network status of a network connection; and, in response to detecting the change in the current network status: adjusting a current value for a predefine flatness threshold for detecting a flat area within the respective frame of the image stream in accordance with the change in the current network status of the network connection; identifying one or more flat areas within the respective image frame in accordance with the predefined flatness threshold; and performing, using a predefined filter, denoising processing on the one or more flat areas that have been identified in accordance with the predefined flatness threshold.

A deep neural network-based reconstruction system for accelerated magnetic resonance imaging of vessel walls. The system can comprise a magnetic resonance imaging (MRI) scanner configured to obtain an image of the vessel walls, and a computer having a processor. The processor comprises a first and second subnetwork implemented in a cascade fashion. The first subnetwork comprises a convolutional neural network (CNN) and an output correcting module. The first subnetwork receives the image and transforms the image to a reduced artifact image. The second subnetwork is an identical duplicate of the first network. The second subnetwork boosts an accuracy of the reduced artifact image to generate a visual representation of the vessel walls. A computer display terminal is connected to the processor and is configured to display the visual representation of the vessel walls.

An image processor receives an input image with enhanced dynamic range to be displayed on a target display which has a different dynamic range than a reference display. After optional color transformation (110) and perceptual quantization (115) of the input image, a multiscale mapping process (120) combines linear (125) and non-linear (130) mapping functions to its input to generate first (127) and second (132) output images, wherein the first and second output images may have a different dynamic range than the first image. A frequency transform (135, 140), such as the FFT, is applied to the first and second output images to generate first (137) and second (142) transformed images. An interpolation function (145) is applied to the first and second transformed images to generate an interpolated transformed image (147). An inverse transform function (150) is applied to the interpolated transformed image to generate an output tone-mapped image (152). The output tone-mapped image is post-processed to be displayed on the target display.

In one embodiment, an apparatus comprises a communication interface and a processor. The communication interface is to communicate with a plurality of devices. The processor is to: receive compressed data from a first device, wherein the compressed data is associated with visual data captured by sensor(s); perform a current stage of processing on the compressed data using a current CNN, wherein the current stage of processing corresponds to one of a plurality of processing stages associated with the visual data, and wherein the current CNN corresponds to one of a plurality of CNNs associated with the plurality of processing stages; obtain an output associated with the current stage of processing; determine, based on the output, whether processing associated with the visual data is complete; if the processing is complete, output a result associated with the visual data; if the processing is incomplete, transmit the compressed data to a second device.

There is provided a computerized method and system of controlling a quality measure in a compression quality evaluation system, the method comprising: calculating a grain value indicative of an extent of grain present in an input image, the grain value being calculated based on one or more features characterizing a base image related to the input image; and configuring the quality measure upon a criterion being met by the value, the quality measure being indicative of perceptual quality of a compressed image compressed from the input image. The calculated grain value may be dependent also on further characteristics of the input image, or in the case of a sequence of images, dependent also on the relation between the image and the preceding image.

Provided are a method and device for generating metadata including frequency characteristic information of an image. Pixel values of a current block among blocks divided from the image are converted into frequency coefficients in the frequency domain. A band value of a frequency band corresponding to each of regions of the current block is determined using the frequency coefficients included in the regions of the current block, the regions of the current block being divided to correspond to different frequency bands. Metadata including the frequency characteristic information of the current block is generated based on the determined band values.

A method of classifying motion from video frames involves generating motion frames indicative of changes in pixel values between pairs of frames. The method also involves determining one-dimensional feature values based on the video frames or motion frames, such as the statistical values or linear transformation coefficients. Each one-dimensional feature value may be stored in a buffer, from which additional temporal feature values can be extracted indicative of the change of the one-dimensional feature values across a set of frames. A classifier may receive the one-dimensional feature values and the additional temporal feature values as inputs, and determine the class of motion present in the video frames. Some classes of motion, such as irrelevant motion, may be considered irrelevant to the execution of certain motion-triggered actions, such that the method may involve suppressing the performance of a motion-triggered action based on the determined class of motion.

Noise reduction in images is provided by performing a noise reduction step on blocks of pixels within a video-processing pipeline. The noise reduction step consists of applying a discrete cosine transform (DCT) to the block of pixels, quantizing the resulting DCT coefficients, and performing an inverse of the DCT to the quantized coefficients. The output of that noise reduction step is a block of image pixels similar to the input pixels, but with significantly less image noise. Because the noise reduction step can be performed quickly on small blocks of pixels, the noise reduction can be performed in real-time in a video processing pipeline.