摘要
为了得到靶场测量中需要的大视场全景视频图像,建立了多镜头大视场视频图像自动拼接系统,对该系统所采用的图像匹配、标记重叠区、无缝拼接等算法进行了研究。介绍小波变换和小波包原理,分析了根据小波系数进行匹配的小波变换图像匹配算法。然后,分析了比较旋转标记重叠区算法,即先匹配出多个拟重叠区,旋转后再做匹配计算,进而标记出唯一重叠区。最后,分析了基于小波包变换的图像无缝拼接算法,即对图像做小波包变换,按频率融合小波系数,再根据这些系数恢复图像。实验结果表明:本系统将4个1.1°×0.88°的视场拼接得到一个大约3.3°×1.76°的大视场,拼接图像平滑,无明显裂缝,处理速度可达10 frames/s,基本满足了靶场测量中大视场视频图像的需求。
In order to get the video image with large visual field in range instrumentation, a system of auto-stitching multi-lens image with large visual field is established and its applied algorithms, such as image matching, marking lapped area, seamless stitching and so on, are investigated. The theories of wavelet transformation and wavelet-packet are presented. An image-matching algorithm based on wavelet-transformation by matching wavelet parameters of image is analyzed. Then, the comparing and rotating algorithm to mark the lapped area is analyzed, in which the only lapped area is marked after sub-lapped areas are matched and rotated. Finally, the image seamless stitching algorithm based on wavelet-packet transformation is analyzed. After image waveletTpacket transformation, the wavelet parameters are merged by frequency and then the image is reconstructed using the result. Experimental results indicate that the system gets an image with 3.3°×1.76° large visual field by stitching four images with 1.1°×0.88~ visual field, the stitched image is smooth and seamless ,the speed of the system can reach up to 10 frames/s. These results show that proposed stitching method can satisfy the requirements for range instrumentation in getting image with large visual field.
出处
《光学精密工程》
EI
CAS
CSCD
北大核心
2008年第11期2145-2150,共6页
Optics and Precision Engineering
基金
国家863高技术研究发展计划资助项目(No.2006AA703104)
关键词
大视场
小波变换
视频图像
图像匹配
无缝拼接
靶场测量
large visual field
wavelet transformation
video image
image matching
seamless stitc-hing
range instrumentation