Abstract: Aspects of the present disclosure relate to object-based image capture. Embodiments include identifying a reference point corresponding to an object in an image of a series of images. Embodiments include comparing a position of the reference point in the image to positions of one or more corresponding reference points in one or more previous images in the series of images. Embodiments include determining a total number of images in the series of images. Embodiments include selecting, based on the comparing and the total number of images in the series of images, between: capturing the image; or declining to capture the image.

Abstract: There is provided with a processing apparatus. A data holder holds at least some of data of a plurality of channels in a target layer among a plurality of layers. Each of a plurality of processors performs, in parallel, a product-sum operation using the data of one channel of the target layer and a coefficient corresponding to the target layer. A selector selects whether to perform first processing or second processing on the basis of information specifying processing in the target layer. The first processing includes inputting the data of one channel of the target layer into one of the plurality of processors. The second processing includes inputting the data of one channel of the target layer to the plurality of processors in parallel.

Abstract: The present invention discloses a method for adaptively detecting chessboard sub-pixel level corner points. Adaptive detection of chessboard sub-pixel level corner points is completed by marking position of an initial unit grid on a chessboard, using a homography matrix H calculated by pixel coordinates of four corner points of the initial unit grid in a pixel coordinate system and world coordinates in a world coordinate system to expand outwards, adaptively adjusting size of an iteration window in the process of expanding outwards, and finally spreading to the whole chessboard region.

Abstract: The present disclosure relates to systems, methods, and computer-readable media for selectively identifying image frames from an input video to provide to an image processing model based on camera statistics. For example, systems disclosed herein include receiving an input video and associated camera statistics from a video capturing device. The systems disclosed herein further include identifying select image frames to provide to the image processing model based on the camera statistics and based on an application of the image processing model. The systems disclosed herein further include selectively identifying and providing camera statistics to the image processing model. By selectively providing data to the image processing model based on camera statistics, the systems disclosed herein can leverage capabilities of video capturing devices to significantly reduce the expense of processing resources when utilizing a variety of image processing models.

Abstract: In an embodiment, a method includes receiving a low-light digital image; generating, by at least one processor, a resulting digital image by processing the low-light digital image with an encoder-decoder neural network comprising a plurality of convolutional layers classified into a downsampling stage and an upsampling stage, and a multi-scale context aggregating block configured to aggregate multi-scale context information of the low-light digital image and employed between the downsampling stage and the upsampling stage; and outputting, by the at least one processor, the resulting digital image to an output device. In the upsampling stage, spatial resolution increases by using a bilinear interpolation operation performed before every few convolutional layers to speed up the inference time of the network.

Abstract: Systems and methods generate a filtering function for editing an image with reduced attribute correlation. An image editing system groups training data into bins according to a distribution of a target attribute. For each bin, the system samples a subset of the training data based on a pre-determined target distribution of a set of additional attributes in the training data. The system identifies a direction in the sampled training data corresponding to the distribution of the target attribute to generate a filtering vector for modifying the target attribute in an input image, obtains a latent space representation of an input image, applies the filtering vector to the latent space representation of the input image to generate a filtered latent space representation of the input image, and provides the filtered latent space representation as input to a neural network to generate an output image with a modification to the target attribute.

Abstract: A method and system for generating a fusion image. The method may include obtaining a first image and a second image of a same scene. Pixels of at least a part of the first image may correspond to pixels of at least a part of the second image. The method may include generating, based on the first image, a first low-frequency image corresponding to a low-frequency component of the first image, the first low-frequency image including a plurality of first pixels. The method may include generating, based on the second image, a first high-frequency image corresponding to a high-frequency component of the second image, the first high-frequency image including a plurality of second pixels, the plurality of second pixels corresponding to the plurality of first pixels.

Abstract: Embodiments are disclosed for video image interpolation. In some embodiments, video image interpolation includes receiving a pair of input images from a digital video, determining, using a neural network, a plurality of spatially varying kernels each corresponding to a pixel of an output image, convolving a first set of spatially varying kernels with a first input image from the pair of input images and a second set of spatially varying kernels with a second input image from the pair of input images to generate filtered images, and generating the output image by performing kernel normalization on the filtered images.

Abstract: Embodiments are generally directed to an adaptive deformable kernel prediction network for image de-noising. An embodiment of a method for de-noising an image by a convolutional neural network implemented on a compute engine, the image including a plurality of pixels, the method comprising: for each of the plurality of pixels of the image, generating a convolutional kernel having a plurality of kernel values for the pixel; generating a plurality of offsets for the pixel respectively corresponding to the plurality of kernel values, each of the plurality of offsets to indicate a deviation from a pixel position of the pixel; determining a plurality of deviated pixel positions based on the pixel position of the pixel and the plurality of offsets; and filtering the pixel with the convolutional kernel and pixel values of the plurality of deviated pixel positions to obtain a de-noised pixel.

Abstract: Provided is an image processing apparatus for generating a high-resolution image. The image processing apparatus includes a memory storing one or more instructions and a processor configured to execute the one or more instructions stored in the memory, wherein the processor is configured to extract feature information regarding a low-resolution image of a current frame by using a first convolutional neural network, generate, based on the feature information, a first high-resolution image of the current frame, remove flickering of the first high-resolution image by using a high-resolution image of a previous frame, and remove flickering of a high-resolution image of a next frame by using at least one of a flickering-removed second high-resolution image of the current frame, or the feature information.

Abstract: A system (116) includes an unlogger (202) configured to unlog logged data, to produce unlogged clipped data. The logged data includes attenuation line integrals and clipping-induced bias. The system further includes a mean estimator (204) configured to estimate a mean value of the unlogged clipped data. The system further includes a correction determiner (206) configured to determine correction to the clipping-induced bias based on the estimated mean value of the unlogged clipped data. The system further includes an adder (210) configured to correct the logged data with the correction to produce corrected logged data.

Abstract: Disclosed herein are an apparatus and method for sensing an image based on an event. The apparatus includes memory in which at least one program is recorded and a processor for executing the program. The program may perform acquiring at least one of brightness information and color information from an input image signal, performing conversion including at least one of filtering of at least one of the acquired brightness information and color information, color conversion, and brightness conversion, calculating a quantized difference between a first converted image converted from the currently input image signal and a second converted image converted from a previously input image signal, and generating event information for the input image signal as a bitstream based on the quantized difference.

Abstract: The electronic device includes a memory storing one or more instructions, and a processor configured to execute the one or more instructions to: receive a selection of a location of a synthesis region in an input image, obtain a target object image to be located in the synthesis region of the input image, by using at least one object detected in the input image, generate a composite image by combining the input image with the target object image, and control a display of the electronic device to display the composite image.

Abstract: A method includes training a first model to measure the banding artefacts, training a second model to deband the image, and generating a debanded image for the image using the second model. Training the first model can include selecting a first set of first training images, generating a banding edge map for a first training image, where the map includes weights that emphasize banding edges and de-emphasize true edges in the first training image, and using the map and a luminance plane of the first training image as input to the first model. Training the second model can include selecting a second set of second training images, generating a debanded training image for a second training image, generating a banding score for the debanded training image using the first model, and using the banding score in a loss function used in training the second model.

Abstract: A system for patient positioning is provided. The system may acquire image data relating to a patient holding a posture and a plurality of patient models. Each patient model may represent a reference patient holding a reference posture, and include at least one reference interest point of the referent patient and a reference representation of the reference posture. The system may also identify at least one interest point of the patient from the image data using an interest point detection model. The system may further determine a representation of the posture of the patient based on a comparison between the at least one interest point of the patient and the at least one reference interest point in each of the plurality of patient models.

Abstract: The present disclosure provides a saliency prediction method and system for a 360-degree image based on a graph convolutional neural network. The method includes: firstly, constructing a spherical graph signal of an image of an equidistant rectangular projection format by using a geodesic icosahedron composition method; then inputting the spherical graph signal into the proposed graph convolutional neural network for feature extraction and generation of a spherical saliency graph signal; and then reconstructing the spherical saliency graph signal into a saliency map of an equidistant rectangular projection format by using a proposed spherical crown based interpolation algorithm. The present disclosure further proposes a KL divergence loss function with sparse consistency. The method can achieve excellent saliency prediction performance subjectively and objectively, and is superior to an existing method in computational complexity.

Abstract: A mobile device that includes a camera and an extended reality software application program is employed by a user in an operating environment, such as an industrial environment. One or more objects within a geofence may be identified. A device crosses within the geofence and acquires sensor data associated with an object within the geofence. The sensor data may include image data and/or audio data. The device or a server system may then determine an object identifier associated with the object based on a comparison of the sensor data with data associated with object identifiers corresponding to objects within the geofence. Based on the object identifier, data associated with the object are obtained. The data associated with the object may be presented via the device, such as an extended reality overlay over a view of the object in the device.

Abstract: Pixel depth information is used to determine a weight to apply to neighboring pixels when using a sharpening filter. A difference between neighboring pixel depths is evaluated and pixels with pixel depths that exceed a threshold are given less weight than other pixels. A sharpening mask may be generated using adjusted pixel colors.

Abstract: Provided is an image processing apparatus for generating a high-resolution image. The image processing apparatus includes a memory storing one or more instructions and a processor configured to execute the one or more instructions stored in the memory, wherein the processor is configured to extract feature information regarding a low-resolution image of a current frame by using a first convolutional neural network, generate, based on the feature information, a first high-resolution image of the current frame, remove flickering of the first high-resolution image by using a high-resolution image of a previous frame, and remove flickering of a high-resolution image of a next frame by using at least one of a flickering-removed second high-resolution image of the current frame, or the feature information.

Abstract: An image processor comprising an input module for receiving image data from an image sensor; an image processing module arranged to perform one or more operations on at least a portion of the image data to generate processed image data; and a characteristic processing module arranged to perform one or more characteristic processing operations on at least a portion of the characteristic data to generate processed characteristic data. The portion of the characteristic data is associated with the portion of the image data; the one or more characteristic processing operations are associated with the one or more image processing operations. The image processor further comprises an output module for outputting the processed image data and processed characteristic data.