In-Vehicle Camera Backlight Compensation by Luminance Blocks
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional methods for correcting backlight images in vehicles with wide-angle cameras are inadequate, leading to reduced recognition performance and visibility issues under diverse backlight conditions, particularly affecting autonomous driving and parking systems.
Innovation Solution
An apparatus and method that segment regions of interest, estimate the horizon position, partition the ROI into blocks, measure average luminance, and adjust luminance values based on backlight conditions to improve solar light source recognition and perform brightness correction.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Measurement precision
If conventional software-based backlight correction methods (such as histogram equalization) are applied globally or in ROI, then the processing covers the entire image or specified regions, but the correction performance deteriorates when the backlight image itself has problems and cannot accurately recognize solar light sources
Solution Approach 1:
The patent divides the image into multiple blocks and further segments each block into sub-blocks to locally analyze luminance characteristics. This segmentation enables precise identification of solar light sources in specific regions without requiring complex global processing, thereby improving recognition accuracy while maintaining manageable complexity.
Solution Approach 2:
The patent applies different processing strategies to different regions of the image based on their backlight conditions. By measuring average luminance values of blocks and identifying regions with abnormal luminance characteristics, the system applies targeted correction only where needed, improving solar light source recognition without unnecessarily complicating the overall processing pipeline.
2Area of stationary object
If wide-angle cameras are used to capture diverse backlight conditions, then the field of view is expanded, but the recognition performance deteriorates due to distorted brightness in white holes and reduced visibility in dark areas
Solution Approach 1:
The patent segments the wide-angle image into multiple blocks, allowing independent analysis of each region's backlight characteristics. This enables the system to handle diverse backlight conditions across the expanded field of view by treating each block according to its specific luminance profile, thereby maintaining reliable recognition performance despite the wide angle distortion.
Solution Approach 2:
The patent focuses correction efforts on specific blocks identified as having backlight problems rather than applying uniform processing to the entire wide-angle image. By identifying blocks with abnormal average luminance values and applying targeted correction, the system maintains recognition reliability in critical areas without wasting resources on regions that don't require correction.
3Productivity
If histogram equalization is applied unconditionally across the global domain, then the processing is simple and fast, but the correction performance deteriorates when there is a problem with the backlight image itself
Solution Approach 1:
The patent performs preliminary analysis by measuring the average luminance value of each block before applying correction. This preliminary step identifies which blocks actually require backlight correction, allowing the system to skip unnecessary processing in normal blocks and apply targeted correction only where needed, thereby maintaining both speed and accuracy.
Solution Approach 2:
The patent applies different correction strategies to different blocks based on their measured luminance characteristics. Blocks identified as having backlight problems receive targeted correction, while normal blocks are left unchanged or receive minimal processing, improving overall correction accuracy without significantly increasing processing time.
Data Source
AI summary
A backlight compensation apparatus includes at least one memory; and at least one processor. By executing instructions, the at least one processor is configured to: receive images from a camera by an in-vehicle system, estimate a position of horizon in the images using intrinsic and extrinsic parameters of the camera, set a region of interest (ROI) based on the position of the horizon, partition the ROI into a plurality of blocks and measure an average luminance value for each of the plurality of blocks, group the blocks into a plurality of block groups based on the average luminance value of each block, determine a backlight condition of the ROI according to a predetermined average luminance value for each block among the plurality of block groups, and adjust a plurality of block luminance values according to the backlight condition of the ROI.