Integrating cameras with hardware is essential for real-time vision in applications like robotics and surveillance. Traditional approaches often face latency and synchronization issues. To overcome these challenges, we propose designing an efficient MIPI camera interface core for low-latency, high-resolution video streaming on an FPGA. The OV5640 image sensor, with a 5MP resolution and up to 3-lane MIPI output, captures 1920×1080 video at 30 fps. Our implementation targets the Digilent MIPI Pcam 5C module and Zybo Z7-20 FPGA board, where video data is stored in DDR memory and output via HDMI. Due to FPGA parallelism,it enables real-time processing for computer vision tasks. The system is applied to a rice sorting machine, where the camera captures images, the FPGA processes them, and the display highlights detected rice in green and shows the count.
Tools: Xilinx Vivado and SDK
Department: Department of Electrical Engineering
Project Poster
