Domain adaptive dehaing based on physical properties

Abstract

To address the issue oflow real-time fusion efficiency in multi-source sensor data, this paper proposes an FPGA-based radar-vision fusion method using D-S evidence theory. By implementing a parallel architecture and pipeline optimization, the fusion latency is reduced to the microsecond level while improving resource utilization. Simulation experiments demonstrate that the proposed method offers high reliability and low latency in autonomous driving scenarios, with potential scalability to intelligent transportation systems.