Research and implementation of large field image real-time mosaic technology based on FPGA
-
摘要: 鉴于目前普遍采用软件方法获得大视场视频图像操作不便且实时性受限的缺陷,研究并设计了一种基于FPGA的可编程技术来实现多个摄像头视频数据实时拼接的大视场成像系统。系统通过APTINA公司的彩色CMOS图像传感器MT9M034获取原始视频图像信息,以Xilinx公司的Virtex-5系列FPGA为核心完成视频数据的实时采集、缓存、拼接及传输。图像拼接部分首先对原始图像进行亮度差异自动调节的预处理以提高整体的拼接效果,然后利用相位相关法完成相对平移量信息检测,对原始图像进行配准,最后采用线性加权融合算法对相邻两幅图像的重合区域进行融合处理,使拼接之后的大视场图像达到渐进渐出平滑过渡的效果。实验结果表明,该成像系统简单可靠,有效地增大了可观测视场,经过拼接处理之后的大视场视频图像清晰度高、实时性强,具有一定的代表性和实用性。Abstract: As a general rule, the software method is used to obtain large field images, which is not timely and convenient. In order to tackle the disadvantage of this method, based on FPGA, a kind of programmable technologies, a large field of view imaging system had put forward and achieved, which can fulfill the real time stitching of the data from multiple cameras. Through the APTINA's color CMOS image sensor MT9M034, the original image information had gained and then the real time data collection, data cache, stitching and transmission had accomplished centering on the Xilinx's Virtex-5 FPGA. Firstly, the automatic adjustment of brightness differences of the original images was preprocessed in order to improve the overall mosaic effect. Secondly, information detection of relative shift amount was completed by the use of phase correlation method to register the original images. Finally, the two adjacent images' overlap area was fused by the use of linear weighted fusion algorithm to make the mosaic image achieve a smoothly fading in and out transitional effect. Experimental results show that the imaging system is simple and reliable, and can effectively increase the field of view of observation. The stitched large field images are of high-definition and real time, with a certain degree of representativeness and practicality.
-
Key words:
- FPGA /
- large field of view imaging /
- real-time mosaic /
- phase correlation /
- linear weighted fusion
-
[1] [2] Huang Qiongdan, Qiu Yuehong, Tian Xiaoping. Image automatic mosaic method based on feature points and optimization theory[J]. Acta Photonica Sinica, 2009, 38(8): 2139-2143. (in Chinese) 黄琼丹, 邱跃洪, 田小平. 基于特征点及优化理论的图像自动拼接方法[J]. 光子学报, 2009, 38(8): 2139-2143. [3] Peng Bo, He Bin. Application and realization of FPGA in video mosaicing[J]. Computer Engineering and Design, 2013, 34(5): 1635-1639. (in Chinese) 彭勃, 何宾. FPGA在视频拼接中的应用与实现[J]. 计算机工程与设计, 2013, 34(5): 1635-1639. [4] [5] Dong Linli. Design of digital image mosaic system based on FPGA[D]. Chengdu: Chengdu University of Technology, 2012. (in Chinese) 董林粒. 基于FPGA的数字图像拼接器系统设计[D]. 成都: 成都理工大学, 2012. [6] [7] [8] MT9M034 Data Sheet. Aptina Imaging[M]. 2010. [9] [10] An Jianni, Liu Guixi. Image mosaic algorithm based on feature points matching and Automatic transform parameters identifying[J]. Infrared and Laser Engineering, 2011, 40(3): 564-569. (in Chinese) 安建妮, 刘贵喜. 利用特征点配准和变换参数自动识别的图像拼接算法[J]. 红外与激光工程, 2011, 40(3): 564-569. [11] Kuglin C, Hines D. The phase correlation image alignment method[C]//Conference on Cybernetics and Society, IEEE, 1975: 163-165. [12] [13] Wu Xianxiang, Guo Baolong, Wang Juan. An improved automatic image sequencing algorithm[J]. Journal of Optoelectronics Laser, 2009, 20(8): 1114-1117. (in Chinese) 吴宪祥, 郭宝龙, 王娟. 一种改进的序列图像自动排序算法[J]. 光电子 激光, 2009, 20(8): 1114-1117. [14] [15] [16] Yang Lei, Cao Jianzhong, Wang Hao, et al. Multifocus image fusion algorithm based on the contourlet transform[C]//SPIE, AOMATT 2012, 84201F: 1-6. [17] Shang Mingshu. A fast image mosaic algorithm based on improved SIFT feature matching[J]. Microelectronics Computer, 2014, 31(1): 64-67. (in Chinese) 尚明姝. 基于改进SIFT特征匹配的快速图像拼接算法[J]. 微电子学与计算机, 2014, 31(1): 64-67. [18] [19] Li Liang, Gu Guohua, Qian Weixian, et al. Infrared image sequence mosaic based on feature points and Poisson fusion[J]. Infrared and Laser Engineering, 2013, 42(9): 2584-2588. (in Chinese) 李亮, 顾国华, 钱惟贤, 等. 基于特征点和泊松融合的红外序列图像拼接[J]. 红外与激光工程, 2013, 42(9): 2584-2588. -
点击查看大图
计量
- 文章访问数: 646
- HTML全文浏览量: 82
- PDF下载量: 915
- 被引次数: 0
下载:
