The present disclosure relates to image processing. The method includes acquiring at least one of a backward propagation feature of an (x+1)th video frame in a video segment or a forward propagation feature of an (x−1)th video frame in the video segment. The video segment includes N video frames, N being an integer greater than 2, and x being an integer. The method further includes deriving a reconstruction feature of the xth video frame from at least one of the xth video frame, the backward propagation feature of the (x+1)th video frame, or the forward propagation feature of the (x−1)th video frame, and deriving a target video frame corresponding to the xth video frame by reconstructing the xth video frame based on the reconstruction feature of the xth video frame. The target video frame has resolution higher than that of the xth video frame.

A signal processing device and an image display apparatus including the same are disclosed. The signal processing device includes a scaler configured to scale input images of various resolutions to a first resolution, a resolution enhancement processor configured to perform learning on the input images and to output a residual image of the first resolution, and an image output interface configured to output an output image of the first resolution based on a scaling image from the scaler and the residual image from the resolution enhancement processor, and the image output interface changes a weight and an application strength of the residual image according to the area of the input image.

An imaging device is presented for use in an imaging system capable of improving the image quality. The imaging device has one or more optical systems defining an effective aperture of the imaging device. The imaging device comprises a lens system having an algebraic representation matrix of a diagonalized form defining a first Condition Number, and a phase encoder utility adapted to effect a second Condition Number of an algebraic representation matrix of the imaging device, smaller than said first Condition Number of the lens system.

A high-resolution microendoscope system includes a light source, a fiber optic bundle configured to transmit light from the light source to a sample, a disc configured to receive light returned from the sample, the disc having spaced apart segments, the spaced-apart segments being at least one of openings and transparent portions, a first camera configured to capture a first image based at least in part on light passing through the disc, and a second camera configured to capture a second image based at least in part on light reflected from the disc.

This application discloses a photographing method and apparatus, to overcome blurring that occurs during photographing. When a zoom ratio is greater than a first zoom ratio threshold, a long-focus camera is started to capture an image. A zoom ratio of the long-focus camera is greater than or equal to the first zoom ratio threshold. Because a high zoom ratio causes large shake, a rotational blur occurs in the image. According to the photographing method disclosed in this application, a first neural network model for rotational image deblurring is used to implement rotational image deblurring processing. In this way, high imaging quality of an image, a video, or a preview image is presented to a user to some extent, and the imaging effect may not be inferior to the effect of photographing with a tripod.

Neural network based systems, methods, and instrumentalities may be used to remove motion artifacts from magnetic resonance (MR) images. Such a neural network based system may be trained to perform the motion artifact removal tasks without reference (e.g., without using paired motion-contaminated and motion-free MR images). Various training techniques are described herein including one that feeds the neural network with pairs of MR images with different levels of motion contamination and forces the neural network learn to correct the motion contamination by transforming a first image of a contaminated pair into a second image of the contaminated pair. Other neural network training techniques are also described with an aim to reduce the reliance on training data that is difficult to obtain.

The present disclosure generally relates to user interfaces for altering visual media. In some embodiments, user interfaces capturing visual media (e.g., via a synthetic depth-of-field effect), playing back visual media (e.g., via a synthetic depth-of-field effect), editing visual media (e.g., that has a synthetic depth-of-field effect applied), and/or managing media capture.

In some implementations, a method includes: determining a complexity value for first image data associated with of a physical environment that corresponds to a first time period; determining an estimated composite setup time based on the complexity value for the first image data and virtual content for compositing with the first image data; in accordance with a determination that the estimated composite setup time exceeds the threshold time: forgoing rendering the virtual content from the perspective that corresponds to the camera pose of the device relative to the physical environment during the first time period; and compositing a previous render of the virtual content for a previous time period with the first image data to generate the graphical environment for the first time period.

A method of de-smearing an image includes capturing image data from an imaging sensor and collecting motion data indicative of motion of the sensor while capturing the image data. The motion data is collected at a higher frequency than an exposure frequency at which the image data is captured. The method includes modeling motion of the sensor based on the motion data, wherein motion is modeled at the higher frequency than the exposure frequency. The method also includes modeling optical blur for the image data, modeling noise for the image data, and forming a de-smeared image as a function of the modeled motion, the modeled blur, and the modeled noise, and the image data captured from the imaging sensor.

Systems/techniques that facilitate self-supervised deblurring are provided. In various embodiments, a system can access an input image generated by an imaging device. In various aspects, the system can train, in a self-supervised manner based on a point spread function of the imaging device, a machine learning model to deblur the input image. More specifically, the system can append to the model one or more non-trainable convolution layers having a blur kernel that is based on the point spread function of the imaging device. In various aspects, the system can feed the input image to the model, the model can generate a first output image based on the input image, the one or more non-trainable convolution layers can generate a second output image by convolving the first output image with the blur kernel, and the system can update parameters of the model based on a difference between the input image and the second output image.