摘要
基准图和实时图公共信息—公因子的提取是提高景象匹配准确率和定位精度的关键因——素。通过对景象匹配误差原因的分析,提出一种有效提取基准图和实时图公因子的方法。该方法首先通过二阶高斯低通滤波算子将图像中同类景象区域化,其后通过Sobel梯度模算子,对图像中较稳定的区域边缘和区域内部轮廓线进行加权,同时滤除不稳定的局部直流分量影响,从而有效提取了基准图和实时图公因子。相关峰分析和多组真实图像实验证明了该方法的正确性和有效性。
The common parts of the reference and real time images,common factor is one of the most important factors which improves the accuracy rate and positing precision of scene match. Through the analysis of scene matching error, a new method was advanced, which can efficiently extract the common factor of the reference and real time images. Firstly, the similar scenes were divided into regions by means of 2-rank Gauss low-pass filter operators,then the steady contour line locating at the region edge and region interior were weighted by means of Sobel gradient operator,meantime,the influences of local direct current components were got rid of, therefore, the common factor of reference and real- time images could be extracted efficiently.The correlative mountain analysis and many real images experiments have proved that this method is correctness and validity.
出处
《红外与激光工程》
EI
CSCD
北大核心
2009年第2期368-372,共5页
Infrared and Laser Engineering
基金
国防预研基金资助项目
关键词
公因子
轮廓线
景象匹配
视觉导航
Common factor
Contour line
Scene matching
Vision-based navigation