Neural network supported camera image or video processing pipelines
AssigneeSpectrum Optix Inc.
InventorsKevin Gordon, Colin D'Amore, Martin Humphreys
Life of this patent
Estimating the term…
Abstract
An image processing pipeline including a still or video camera includes a first neural network arranged to process and provide a neural network based result for at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing. A second neural network is arranged to receive the neural network result and further provide at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing, wherein at least one of the first and second neural networks generate data on local processors supported by the still or video camera.
Claims17
An image processing pipeline system including a still or video camera, comprising:
- a first neural network arranged to process and provide a neural network based result for at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing;
- and
- a second neural network arranged to receive the first neural network result and further provide at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing, wherein at least one of the first and second neural networks generate data on local processors supported by the still or video camera, wherein at least one of the first neural network and the second neural network is calibrated using calibration data from one or more targets for assessing any combination of resolution, focus, and dynamic range, and raw profiling data that includes natural and manmade scenes captured from a reference system and procedurally generated scenes that are mathematically derived.
An image processing pipeline system including a still or video camera, comprising:
- using a first neural network arranged to process and provide a neural network based result for at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing;
- and
- arranging a second neural network to receive the first neural network result and further provide at least one of an image capture setting, sensor processing, global post processing, local post processing, and portfolio post processing, wherein at least one of the first and second neural networks generate data on local processors supported by the still or video camera, wherein at least one of the first neural network and the second neural network is calibrated using calibration data from one or more targets for assessing any combination of resolution, focus, and dynamic range, and raw profiling data that includes natural and manmade scenes captured from a reference system and procedurally generated scenes that are mathematically derived.
An image processing pipeline method, comprising:
- processing data using a first neural network to provide an image capture setting for a still or video camera,
- using a second neural network to provide sensor processing for an image captured using image capture settings provided by the first neural network,
- using a third neural network to provide sensor global post processing for a sensor processed image provided by the second neural network,
- using a fourth neural network to provide local post processing for a globally post processed image provided by the third neural network;
- and
- using a fifth neural network to provide portfolio post processing for a locally post processed image provided by the fourth neural network, wherein at least one neural network is calibrated using calibration data from one or more targets for assessing any combination of resolution, focus, and dynamic range, and raw profiling data that includes natural and manmade scenes captured from a reference system and procedurally generated scenes that are mathematically derived.
An image capture device, comprising:
- at least one processor to control image capture device operation;
- and
- at least one neural processor supported by the image capture device and connected to the processor to receive neural network data, with the neural processor using neural network data to provide at least two separate neural network processing procedures selected from a group including sensor processing, global post processing, and local post processing, wherein the neural processor supports at least one neural network calibrated using calibration data from targets for assessing any combination of resolution, focus, and dynamic range, and raw profiling data that includes natural and manmade scenes captured from a reference system and procedurally generated scenes that are mathematically derived.
Dependencies are inferred from the claim text.