Systems, methods, and media for generating digital images using low bit depth image sensor data

A method for generating digital images having a bit depth higher than an image sensor bit depth sensed by an image sensor, the method comprising:

• receiving, from a communication system, a series of image frames, wherein the series of image frames are output from the image sensor for a given scene and comprise data at the image sensor bit depth for a plurality of pixels;
• providing data, based on the image sensor bit depth, from a plurality of image frames of the series of image frames to a trained machine learning model, wherein the trained machine learning model:
• processes spatiotemporal relationships among pixels across the plurality of image frames, and predicts pixel intensity values based on learned representations of the plurality of image frames;
• and
• generating an enhanced image of the given scene based on an output of the trained machine learning model, the enhanced image having the higher bit depth.

A non-transitory computer readable medium containing computer executable instructions that, when executed by a processor, cause the processor to perform a method for generating digital images, comprising:

• receiving, from a communication system, a series of image frames, wherein the series of image frames are output from an image sensor for a given scene and comprise data at an image sensor bit depth for a plurality of pixels and a first noise content;
• providing data, based on the image sensor bit depth, from a plurality of initial image frames of the series of image frames to a trained machine learning model, wherein the trained machine learning model identifies spatial and temporal patterns of pixels across the plurality of initial image frames, and learns statistical priors of the initial image frames to predict pixel intensity values;
• and
• generating a plurality of enhanced image frames of a scene based on an output of the trained machine learning model, wherein the plurality of enhanced image frames comprises a bit depth higher than the image sensor bit depth and a noise content lower than the first noise content.