Vehicle Camera Low-Light Image Correction for ADAS Control
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Advanced Driver Assistance Systems (ADAS) face challenges in processing low-illuminance images due to the lower performance of software image signal processors (ISPs) used in vehicles, leading to degraded image quality and performance of applications like object detection and surround view monitoring.
Innovation Solution
A driving assistance apparatus and method that employs an artificial neural network to recognize low-illuminance environments and generate corrected image data, improving image quality by processing image data from cameras and using an illuminance sensor to determine the need for correction, which is then used for vehicle control.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Manufacturing precision
If a high-performance image signal processor (ISP) is used to improve image quality in low-illuminance environments, then image quality is improved, but cost, power consumption, and space requirements increase
Solution Approach 1:
The patent replaces the hardware-based high-performance ISP with a software-based neural network model running on an embedded board. This substitution transforms the image processing approach from a complex hardware system to a software solution, maintaining image quality improvement while reducing cost, power consumption, and space requirements.
Solution Approach 2:
The patent changes the operating parameters of the image processing system by using a neural network model that can dynamically adjust processing parameters based on illuminance conditions. The controller determines whether to apply the neural network model based on illuminance sensor data, optimizing performance for low-illuminance environments while avoiding unnecessary processing in well-lit conditions.
2Device complexity
If a software ISP is used to reduce cost, power consumption, and space, then device complexity is reduced, but image quality in low-illuminance environments deteriorates
Solution Approach 1:
The patent introduces an illuminance sensor as an intermediary component that detects lighting conditions and triggers the neural network model when low-illuminance is detected. This intermediary enables the simple software ISP to achieve high-performance image processing selectively, maintaining low cost and power consumption while improving image quality when needed.
Solution Approach 2:
The patent performs preliminary action by training the neural network model in advance with low-illuminance image data. The model is pre-trained to handle low-illuminance conditions, so when such conditions are detected during operation, the model can immediately process images effectively without requiring complex real-time adjustments.
3Speed
If image processing is performed without neural network correction in low-illuminance environments, then processing speed is maintained, but application function performance (object detection, surround view monitoring) deteriorates
Solution Approach 1:
The patent implements dynamic image processing by having the controller adjust processing based on illuminance conditions. The system dynamically switches between standard processing and neural network-based processing, ensuring that application functions receive high-quality images when needed while maintaining efficient processing during normal conditions.
Data Source
AI summary
A driving assistance apparatus includes a camera disposed on at least one of a front, sides, or a rear of a vehicle and configured to capture an image of at least one of the front, sides, or rear of the vehicle, and a controller configured to process image data provided from the camera, wherein the controller recognizes whether a surrounding environment of the vehicle is a low-illuminance environment, generates corrected image data using a first artificial neural network model trained to generate the corrected image data in which the image data has been corrected based on the image data when recognizing the low-illuminance environment, and performs traveling control of the vehicle based on the generated corrected image data.