摘要
为了解决当前图像拼接算法在背景复杂多变和重叠区域少的情况下,导致算法拼接不准确的问题,论文分别从角点检测与特征配准的角度出发,提出了基于角点检测与特征点配准的图像拼接算法。首先,根据图像Hessian矩阵,构建尺度空间,通过积分图像转换,设计基于surf的角点检测算子,精确定位角点,达到精准定位图像拼接参考点的目的。然后,根据角点匹配参数,推导计算出单应性矩阵,进行两幅图像间拼接转换处理,进一步精确并全自动化图像拼接结果。最后,基于软件开发环境VS2015实现算法,并系统集成。实验测试结果显示:与当前图像拼接技术相比,论文算法拥有更高的准确性与稳定性。
In order to solve the current image stitching algorithm in complex background and overlapping area of less,cause the algorithm of joining together the problem of inaccurate,this paper respectively from the perspective of corner detection and image registration,is proposed based on corner detection and image matching algorithm of image registration.First of all,according to the image Hessian matrix,build the scale space,through the integral image conversion,accurate positioning angular point,to achieve the purpose of precise image mosaicing reference point.Then,according to the angular point matching parameters,homography calculate the single matrix is derived,splicing transformation between the two images processing,accurate and automatic image matching results further.Finally,based on the software development environment VS2015 algorithm,and system integration.Test results show that compared with the current image splicing technology,this algorithm has higher accuracy and stability.
作者
李养胜
李俊
LI Yangsheng;LI Jun(Shaanxi Industrial Vocational College,Xianyang 712000)
出处
《舰船电子工程》
2018年第4期69-72,共4页
Ship Electronic Engineering
基金
陕西工业职业技术学院科研计划项目(编号:zk14-24
zk16-14)资助
