HDR Tone Mapping Pipeline for Autonomous Vehicle Vision
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Autonomous vehicles face challenges in efficiently processing high dynamic range (HDR) images due to limited data transfer bandwidth and processing power, leading to inefficiencies in object detection and potential safety delays, as existing tone mapping methods require separate workflows for online and offline processing, consuming additional computational resources and causing delays.
Innovation Solution
A unified approach is implemented where global tone mapping is performed online to reduce HDR images to low dynamic range (LDR) for neural network processing, and local tone mapping is applied offline for human vision, optimizing bit allocation based on pixel values and end-task requirements, thereby reducing data duplication and resource consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If separate workflows are used for online and offline tone mapping processing, then processing accuracy for different tasks is improved, but computational resources are consumed and processing delays occur
Solution Approach 1:
The patent segments the tone mapping process into global tone mapping for online neural network processing and local tone mapping for offline human vision processing. This segmentation allows each workflow to be optimized for its specific requirements while maintaining a unified overall structure, resolving the contradiction between processing accuracy and efficiency.
Solution Approach 2:
The patent implements a unified tone mapping system that serves multiple functions: it provides global tone mapping for online processing and local tone mapping for offline processing within a single integrated framework. This multi-functionality eliminates the need for completely separate workflows, reducing computational resource consumption while maintaining task-specific accuracy.
2Measurement precision
If HDR images are processed directly without tone mapping, then image quality and dynamic range are preserved, but data transfer bandwidth and processing power requirements increase
Solution Approach 1:
The patent applies tone mapping transformations that change the parameter space of image data from high dynamic range to low dynamic range. This parameter transformation reduces the bit depth and computational requirements while preserving essential image quality information, directly addressing the contradiction between image quality and computational resource consumption.
3Speed
If global tone mapping is applied to all images, then processing speed is improved, but image quality for human viewing is degraded
Solution Approach 1:
The patent applies local tone mapping specifically to regions or images that require enhanced quality for human viewing, while using global tone mapping for online neural network processing. This localized application of quality enhancement resolves the contradiction between processing speed and image quality by applying the appropriate treatment to the appropriate data.
Data Source
AI summary
A system is provided that includes an image sensor coupled to a vehicle, and control circuitry configured to perform operations including receiving, from the image sensor, an input stream comprising high dynamic range (HDR) image data associated with an environment of the vehicle, and processing the input stream at the vehicle by applying a global tone mapping, followed by offline image processing that can include applying a local tone mapping to the globally tone mapped images of the same input stream.