Design and Implementation of Multi Exposure Smart Vehicular Camera Applying Auto Exposure Control Algorithm Based on Region of Interest

Abstract

Recently, many researches are carried out for Advanced Driver Assistant Systems(ADAS). Especially, many studies are carried out to analyze the road situation using road images. In order to improve the performance of the road situation analysis, it is necessary to acquire images with appropriate exposure time. In this paper, we design and implement multi exposure smart vehicular camera which provides road traffic information to driver. Proposed device can acquire road traffic information by on-board camera and various sensors. And we propose an auto exposure control algorithm for the road environment to increase accuracy of image recognition. In addition, we also propose the switching ROI method that apply existing ROI techniques to overcome a limited computation power of embedded devices. We developed prototype of multi exposure smart vehicular camera and performed experiments to evaluate proposed auto exposure control algorithm and switching ROI method. The results show that the average accuracy of image recognition increased by 13.45%.

Personalized Exposure Control Using Adaptive Metering and Reinforcement Learning

“…Techniques for more specialized treatment of frontand back-lit scenes have been described (e.g., [8]), but they fundamentally rely on histogram analysis that does not jointly consider object importance and lighting condition. Some native cameras on popular phones prioritize exposure on detected face regions [9], [10], [11]; many allow user interaction to select a region of interest for appropriate exposure adjustment. In contrast, our auto-exposure approach supports more general scenes beyond faces, and it does not require user interaction.…”