Person Backlighting for Improved AEC and AGC
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Video cameras, especially panoramic ones, face challenges in capturing optimal images of multiple people due to insufficient dynamic range and inadequate backlighting, leading to under or overexposure, which standard methods struggle to address effectively.
Innovation Solution
The system determines regions of interest using image motion, sound source localization, and active speaker detection, followed by automatic exposure and gain control adjustments, enhancing image quality by selecting and focusing on specific individuals or objects within the scene.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Area of stationary object
If standard backlighting is used in panoramic cameras, then the camera can capture multiple people in the scene, but the image quality of individual people becomes insufficient due to insufficient dynamic range and inadequate backlighting
Solution Approach 1:
The patent divides the scene into multiple regions of interest, each corresponding to a detected person or object. Instead of applying uniform backlighting to the entire panoramic scene, the system segments the image and applies selective exposure and gain control to specific regions, thereby maintaining image quality across multiple subjects without compromising overall coverage.
Solution Approach 2:
The patent implements local quality enhancement by applying different exposure and gain parameters to different regions of the panoramic image. Each region of interest receives customized backlighting adjustments based on its specific lighting conditions and importance, rather than using a single global setting for the entire scene.
2Manufacturing precision
If face detection is used to dynamically change the backlighting region of interest, then image quality can be improved for detected faces, but the method fails when image quality or facial resolution is insufficient for face recognition
Solution Approach 1:
The patent employs multiple detection methods including face detection, object detection, and motion detection to identify regions of interest. This multi-functional approach ensures that the system can operate effectively regardless of whether face recognition is feasible, making the backlighting adjustment mechanism universally applicable across various scenarios including low-resolution or non-face subjects.
Solution Approach 2:
The patent introduces intermediate detection mechanisms such as motion detection and general object detection that serve as fallback or complementary methods when face detection fails. These intermediary detection methods enable the system to identify regions of interest even when facial features are not clearly visible or recognizable.
3Manufacturing precision
If automatic exposure and gain control adjustment is applied to selected regions, then image quality of regions of interest is improved, but the system complexity increases due to multiple detection and control mechanisms
Solution Approach 1:
The patent implements dynamic region of interest selection and exposure control that adapts in real-time based on detected motion, face positions, and lighting conditions. The system continuously updates which regions receive enhanced backlighting adjustments, allowing flexible adaptation to changing scenes without requiring complex manual configuration or fixed processing pipelines.
Data Source
AI summary
Regions of interest in video image capture for communication purposes are selected based on one or more inputs based on sound source localization, multi-person detection, and active speaker detection using audio and/or visual cues. Exposure and/or gain for the selected region are automatically enhanced for improved video quality focusing on people or inanimate objects of interest.