Quanta Image Frame HDR Capture for Low-Light Motion Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing imaging systems face challenges in generating high dynamic range (HDR) images due to issues such as poor signal-to-noise ratio (SNR) caused by low-light conditions or fast motion, leading to failed image registration and ineffective blending of multiple frames.
Innovation Solution
Utilizing a multi-mode image sensor to capture first image frames with a higher bit depth and second image frames with a lower bit depth, where the second frames are captured between consecutive pairs of first frames, and processing these frames through a multi-frame processing pipeline to generate an HDR image.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Illumination intensity
If multiple image frames are captured and blended to produce HDR image, then the dynamic range and image details are improved, but the signal-to-noise ratio deteriorates in low-light or fast-motion conditions
Solution Approach 1:
The patent implements a dynamic frame selection mechanism that adapts to scene conditions. The system determines whether to use single-frame or multi-frame processing based on motion detection and scene stability, dynamically adjusting the processing pipeline to maintain high signal-to-noise ratio while achieving HDR效果 when conditions permit
Solution Approach 2:
The patent changes the parameter of frame selection criteria by introducing scene-based decision-making. Different exposure parameters and blending weights are applied depending on whether the scene is static or dynamic, optimizing both dynamic range and signal-to-noise ratio for different conditions
2Loss of information
If multiple image frames are captured and blended to produce HDR image, then greater image details are incorporated, but image registration fails under certain conditions
Solution Approach 1:
The patent performs preliminary scene analysis and motion detection before committing to multi-frame processing. By pre-assessing scene stability and motion levels, the system prepares appropriate registration strategies in advance, preventing registration failures before they occur
Solution Approach 2:
The patent introduces scene-based decision logic as an intermediary between frame capture and blending. This intermediary layer analyzes scene conditions and mediates the processing pipeline selection, ensuring that multi-frame blending is only applied when registration is likely to succeed
3Illumination intensity
If multiple image frames are captured at different exposure levels, then the blending process improves HDR quality, but the processing time increases
Solution Approach 1:
The patent applies partial multi-frame processing by selecting only the necessary number of frames based on scene conditions. Instead of always processing all captured frames, the system processes only the optimal subset required to achieve HDR quality, reducing unnecessary processing time
Solution Approach 2:
The patent implements dynamic processing depth adjustment where the extent of multi-frame processing adapts to scene requirements. Static scenes receive full multi-frame processing for maximum HDR quality, while dynamic scenes receive reduced processing to minimize latency and processing time
Data Source
AI summary
A method includes obtaining, using a multi-mode image sensor, first image frames and second image frames. The first image frames include image data having a first bit depth, and the second image frames include image data having a second bit depth smaller than the first bit depth. The first image frames are captured using two or more exposure times, and the second image frames are captured using at least one exposure time shorter than the two or more exposure times. At least one of the second image frames is captured in between each consecutive pair of the first image frames. The method also includes providing at least the first image frames to a multi-frame processing (MFP) pipeline and processing at least the first image frames using the MFP pipeline to generate an output image.