TOPIC

• Titile: High Dynamic Range Object Detection System with Image Fusion Network using High-Illumination Specialized Binary Image

INFORMATION

• Producer: Hyeong-Ung Byeon
• Published: IEEE Sensors Conference'24
• Paper: Link

IMAGE AND VIDEO

OVERVIEW

• This study presents an object detection system combining a dual-imaging CMOS image sensor (CIS) with a fusion model to address feature loss from light saturation. The CIS captures general and binary images, with the latter providing additional information in saturated conditions. These images are fused to preserve texture details and enhance detection under high-intensity lighting. Using the PascalRaw dataset, the proposed system demonstrates robust performance across various lighting conditions, significantly outperforming traditional methods. This research highlights the effectiveness of machine-friendly sensors and fusion models in improving object detection accuracy.