Abstract: A face located along a stitch line in a spherical image is detected by rendering views of regions of the spherical image along the stitch line. The spherical image may be produced by combining first and second images. A first view of a projection of the spherical image is rendered. A scaling factor for rendering a second view of the projection is determined based characteristics of the first portion of the face. The second view is then rendered according to the scaling factor. The use of the scaling factor to render the second view causes a change in the depiction of the second portion of the face. For example, the scaling factor can indicate to change the resolution or expected size of the second portion of the face when rendering the second view. A face is then detected within the spherical image based on the rendered first and second views.

Abstract: Systems and methods are disclosed for non-local means denoising of images. For example, methods may include receiving an image from an image sensor; determining a set of non-local means weights for the image; applying the set of non-local means weights to the image to obtain a first denoised image; applying the set of non-local means weights to the first denoised image to obtain a second denoised image; and storing, displaying, or transmitting an output image based on the second denoised image.

Abstract: Systems and methods are disclosed for image signal processing. For example, methods may include receiving an image from an image sensor; applying a filter to the image to obtain a low-frequency component image and a high-frequency component image; determining a first enhanced image based on a weighted sum of the low-frequency component image and the high-frequency component image, where the high-frequency component image is weighted more than the low-frequency component image; determining a second enhanced image based on the first enhanced image and a tone mapping; and storing, displaying, or transmitting an output image based on the second enhanced image.

Abstract: An image capture device may include an image sensor, a processor, and memory. The image sensor may be configured to obtain an image. The processor may be configured to: generate a grid on the image forming tiles; determine a fringing level of each vertex of the vertices; sort all of the tiles based on the fringing level of each tile so that the tiles are sorted in a descending order from the tile with a highest of the fringing levels to the tile with a lowest of the fringing levels; and apply a fringing compensation to a subset of the sorted tiles to obtain a processed image. The memory may be configured to store the processed image.

Abstract: Methods and apparatus for processing of video content to optimize codec bandwidth. In one embodiment, the method includes capturing panoramic imaging content (e.g., a 360° panorama), mapping the panoramic imaging content into an equi-angular cubemap (EAC) format, and splitting the EAC format into segments for transmission to maximize codec bandwidth. In one exemplary embodiment, the EAC segments are transmitted at a different frame rate than the subsequent display rate of the panoramic imaging content. For example, the mapping and frame rate may be chosen to enable the rendering of 8K, 360° content at 24 fps, using commodity encoder hardware and software that nominally supports 4K content at 60 fps.

Abstract: Video content may be captured by an image capture device during a capture duration. The video content may include video frames that define visual content viewable as a function of progress through a progress length of the video content. Rotational position information may characterize rotational positions of the image capture device during the capture duration. Time-lapse video frames may be determined from the video frames of the video content based on a spatiotemporal metric. The spatiotemporal metric may characterize spatial smoothness and temporal regularity of the time-lapse video frames. The spatial smoothness may be determined based on the rotational positions of the image capture device corresponding to the time-lapse video frames, and the temporal regularity may be determined based on moments corresponding to the time-lapse video frames. Time-lapse video content may be generated based on the time-lapse video frames.

Abstract: Flare compensation includes receiving a first image and a second image; converting the first and the second images from an RGB domain to a YUV domain; obtaining an intensity differences profile along a stitch line between the first and the second images, where the intensity differences profile is obtained for the Y component; obtaining a dark corner intensity differences profile between the first and the second images based on a relative illumination of an area outside a first image circle of the first image and a second image circle of the second image, where the dark corner intensity differences profile is obtained for the Y component; obtaining a flare profile using the intensity differences profile and the dark corner intensity differences profile; converting the flare profile of the Y component to an RGB flare profile; and modifying one of the first or second images based on the RGB flare profile.

Abstract: Visual content is captured by an image capture device during a capture duration. The image capture devices experiences change in position during the capture duration. The trajectory of the image capture device is smoothed based on a look ahead of the trajectory. A punchout of the visual content is determined based on the smoothed trajectory. The punchout of the visual content is used to generate stabilized visual content.

Abstract: An image or a video may include a spherical capture of a scene. A punchout of the image or the video may provide a panoramic view of the scene.

Abstract: Visual content is captured by an image capture device during a capture duration. The image capture devices experiences motion during the capture duration. The intentionality of the motion of the image capture device is determined based on angular acceleration of the image capture device during the capture duration. A punchout of the visual content is determined based on the intentionality of the motion of the image capture device. The punchout of the visual content is used to generate stabilized visual content.

Abstract: Methods and apparatus for processing of video content to optimize codec bandwidth. In one embodiment, the method includes capturing panoramic imaging content (e.g., a 360° panorama), mapping the panoramic imaging content into an equi-angular cubemap (EAC) format, and splitting the EAC format into segments for transmission to maximize codec bandwidth. In one exemplary embodiment, the EAC segments are transmitted at a different frame rate than the subsequent display rate of the panoramic imaging content. For example, the mapping and frame rate may be chosen to enable the rendering of 8K, 360-degree content at 24 fps, using commodity encoder hardware and software that nominally supports 4K content at 60 fps.

Abstract: A system accesses an image with each pixel of the image having luminance values each representative of a color component of the pixel. The system generates a first histogram for aggregate luminance values of the image, and accesses a target histogram for the image representative of a desired global image contrast. The system computes a transfer function based on the first histogram and the target histogram such that when the transfer function is applied, a histogram of the modified aggregate luminance values is within a threshold similarity of the target histogram. The system modifies the image by applying the transfer function to the luminance values of the image to produce a tone mapped image, and outputs the modified image.

Abstract: An image capture device may include an image sensor, a processor, and a memory. The image sensor may be configured to obtain an image. The processor may be configured to generate a grid on the image. The grid may include one or more vertices. The one or more vertices may be used to form tiles. The processor may be configured to determine a flare level of each vertex. The processor may be configured to assign a maximum flare level for each tile of the image. The processor may be configured to sort the tiles. The tiles may be sorted based on the maximum flare level of each tile. The processor may be configured to apply a flare compensation to a subset of the tiles to obtain a processed image. The processed image may have reduced flare artifacts or no flare artifacts. The processed image may be stored in the memory.

Abstract: Methods and apparatus for processing of video content to optimize codec bandwidth. In one embodiment, the method includes capturing panoramic imaging content (e.g., a 360° panorama), mapping the panoramic imaging content into an equi-angular cubemap (EAC) format, and splitting the EAC format into segments for transmission to maximize codec bandwidth. In one exemplary embodiment, the EAC segments are transmitted at a different frame rate than the subsequent display rate of the panoramic imaging content. For example, the mapping and frame rate may be chosen to enable the rendering of 8K, 360° content at 24 fps, using commodity encoder hardware and software that nominally supports 4K content at 60 fps.

Abstract: Systems and methods are disclosed for high dynamic rate processing based on angular rate measurements. For example, methods may include receiving a short exposure image that was captured using an image sensor; receiving a long exposure image that was captured using the image sensor; receiving an angular rate measurement captured using an angular rate sensor attached to the image sensor during exposure of the long exposure image; determining, based on the angular rate measurement, whether to apply high dynamic range processing to an image portion of the short exposure image and the long exposure image; and responsive to a determination not to apply high dynamic range processing to the image portion, selecting the image portion of the short exposure image for use as the image portion of an output image and discard the image portion of the long exposure image.

Abstract: Systems and methods are disclosed for image signal processing. For example, systems may include an image sensor and a processing apparatus. The image sensor captures image data using a plurality of selectable exposure times. The processing apparatus receives a first image from the image sensor captured with a first exposure time and receives a second image from the image sensor captured with a second exposure time that is less than the first exposure time. A high dynamic range image is determined based on the first image and the second image, wherein an image portion of the high dynamic range image is based on a corresponding image portion of the second image when a pixel of a corresponding image portion of the first image is saturated. An output image that is based on the high dynamic range image is stored, displayed, or transmitted.

Abstract: Systems and methods are disclosed for image signal processing. For example, methods may include receiving an image from an image sensor; applying a filter to the image to obtain a low-frequency component image and a high-frequency component image; determining a first enhanced image based on a weighted sum of the low-frequency component image and the high-frequency component image, where the high-frequency component image is weighted more than the low-frequency component image; determining a second enhanced image based on the first enhanced image and a tone mapping; and storing, displaying, or transmitting an output image based on the second enhanced image.

Abstract: Visual content is captured by an image capture device during a capture duration. The image capture devices experiences change in position during the capture duration. The trajectory of the image capture device is smoothed based on a look ahead of the trajectory. A punchout of the visual content is determined based on the smoothed trajectory. The punchout of the visual content is used to generate stabilized visual content.

Abstract: An image is processed using a combination of techniques, particularly, in which low frequency information is processed using multiple tone control and high frequency information is processed using local tone mapping. An image is divided into a plurality of blocks including a given block. Low frequency information and high frequency information of the given block are separated. The low frequency information is processed using multiple tone control. The low high frequency information is processed using local tone mapping. A processed image is then produced based on the processed low frequency information and based on the processed high frequency information, the processed image corresponding to the image captured using the image sensor. The processed image is then output for storage or display. Processing the low frequency information can include using a gain curve and bilinear interpolation. Processing the high frequency information can include using an edge preservation filter.

Abstract: Methods and apparatus for shared image processing among multiple devices. In one embodiment, an exemplary action camera performs a partial multiband blend. Even though the action camera may not have resources to handle the multiband blend of the entire action camera's footage, it can do a significant portion. The partially blended content can be used in ready-to-share applications, or completely blended by another device.