基于FPGA的图像拼接技术研究与实现

Research and Design of Image Mosaic Technology Based on FPGA

由于目前成熟的图像拼接技术普遍采用软件方法,获得的大视场视频会有不便于实时性的缺陷,提出了一种基于FPGA可编程技术控制多个CMOS摄像头视频数据并实时拼接的方法。该系统采用双目摄像头OV5640获取两路分辨率为320*240的图像数据,用Altera公司的Cyclone IV系列FPGA芯片作为核心芯片完成对两路CMOS视频图像的采集和处理,并通过两片DDR2芯片完成图像数据的缓存,视频图像经过拼接后由VGA显示。图像拼接部分主要通过以实验为基础的相位标校法完成两路图像的配准工作,采用加权算法进行图像融合工作,消除拼接缝,拼接后的视频分辨率为406*192,视场得到扩大,满足实时性要求,具有一定代表性。

For the moment, mature image Mosaic technology generally adopts software method, which is not timely. In order to solve this problem, in this paper, a method which based on FPGA programmable technology was proposed to control multiple CMOS cameras video data and real time splicing. In this system, the dual camera OV5G40 was used to get two image data which the resolution was 320-240 and the core FPGA chip which belongs to Altera＇s Cyclone IV series to collect and process two video images. Then two DDR2 chips were used to complete the cache of image data. Finally, video image was displayed by VGA after Mosaic. Image splicing part mainly through phase calibration method based on the experiment of two road image registration work, weighted algorithm was adopted to improve the image fusion to eliminate the seam; after joining together the video the resolution was 406.192; video field was expanded to meet the real time requirement, which is representative.