Object-Based Auto Exposure Using Neural Networks for Dynamic Scenes
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional auto exposure (AE) systems in cameras fail to adapt to moving and changing scenes due to the use of fixed weighting tables, limiting image quality and perception in automotive systems.
Innovation Solution
Implementing object-based auto exposure using neural network models that dynamically generate weighting tables based on object detection, considering factors like object type, distance, size, and location to adjust brightness levels in video frames.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Adaptability or versatility
If fixed weighting tables are used in conventional AE systems, then the system structure is simple and easy to implement, but the system cannot adapt to moving objects and changing scenes, limiting image quality and perception ability
Solution Approach 1:
The patent transforms the static fixed weighting tables into dynamic adaptive weighting tables that are generated in real-time based on object detection results. The weighting table is continuously updated according to object characteristics (type, distance, size, location) detected by neural network models, enabling the AE system to adapt to moving objects and changing scenes while maintaining manageable complexity through automated generation processes
Solution Approach 2:
The patent changes the parameters of the weighting table from fixed predetermined values to dynamically adjusted values based on object characteristics. By varying weighting parameters according to detected object properties (distance, size, type, location), the system achieves adaptability without requiring complex manual configuration, as the parameters are automatically modified based on scene analysis
2Extent of automation
If manual manipulation of weighting tables is used, then some adaptability is achieved, but the system cannot automatically adapt to objects moving and changing scenes in real-time
Solution Approach 1:
The patent implements self-service by enabling the AE system to automatically generate and update its own weighting tables based on real-time object detection. The system uses neural network models to detect objects and automatically computes appropriate weighting values without external intervention, achieving full automation while managing complexity through integrated detection and control mechanisms
Solution Approach 2:
The patent establishes a feedback loop where object detection results continuously inform weighting table adjustments. The neural network detects objects and their characteristics, which feed back to the AE module to dynamically modify weighting parameters, creating an automated closed-loop system that adapts to moving objects without manual intervention while maintaining controlled complexity through systematic feedback processing
3Manufacturing precision
If conventional AE systems with fixed weighting tables are used, then the system is computationally simple, but image quality and perception ability for automotive applications are limited
Solution Approach 1:
The patent segments the image processing task by dividing the weighting table into region-specific segments based on detected object locations. Instead of applying a uniform fixed weighting, the system creates distinct weighting regions around detected objects (pedestrians, vehicles, cyclists) with different weight values, allowing precise control of exposure for each segment while maintaining overall system manageability through localized processing
Data Source
AI summary
An apparatus comprising an interface and a processor. The interface may be configured to receive pixel data. The processor may be configured to process the pixel data arranged as video frames, perform computer vision operations on the video frames to detect objects in the video frames, analyze characteristics of the objects detected, determine adaptive auto-exposure weightings in response to the characteristics of the objects detected, generate an auto-exposure weighting table based on the adaptive auto-exposure weightings and generate output video frames. Auto-exposure for the output video frames may be determined in response to the auto-exposure weighting table and the video frames. The auto-exposure weighting table may be generated for each of the video frames.