@

초록

With the advancement of autonomous driving and advanced driver assistance systems (ADAS), the importance of vehicle surround environment recognition technology is growing. Surround view monitors (SVMs) are a technology that synthesizes multiple camera images to provide a 360-degree bird's-eye view, and have become an essential ADAS feature in modern vehicles, used not only for parking and low-speed driving convenience but also for precision autonomous driving. This paper proposes a scalable SVM architecture based on a single Zynq SoC platform. The proposed system dynamically controls calibration parameters and stitching coordinates in the PS, making it easy to change viewpoints and maintain the system. Fisheye correction and bird's-eye view conversion preserve 180-degree field of view information using a reverse mapping method, and real-time quality optimization is performed through camera sensor gamma and white balance control and programmable logic (PL) Gamma LUT-based correction. In addition, by integrating a deep learning processing unit (DPU) within the same chip, the system demonstrates the ability to extend beyond simple viewer functions to include object detection and situation awareness functions.

키워드