摘要
遥感图像中目标的空间定位问题一般是通过模板匹配或者特征匹配的方法来实现的。当给定的目标模板与遥感影像中候选识别的目标存在着空间角度、缩放比例和背景噪声较强等问题时,采取模板匹配方法就很难实现对目标的快速准确定位。遗传算法是一种自适应的迭代寻优搜寻和直接对参数对象进行操作的智能算法。利用这种算法优势可以屏蔽掉模板匹配目标的过程中对复杂参数的确定过程。本文将遗传算法与模板匹配方法密切结合,选取一幅北京奥运规划区遥感影像进行了方法实验,结果表明该方法除具有遗传算法智能快速化的效果,还具有实际操作简便的优点。
Generally, template matching or feature matching is used to implement spatial object localization. While with different object template or object area with rotation,scaling and strong background noise in remote sensing image,it is difficult to make sure in the traditional method. Genetic algorithm is an intelligent method possessing adaptive stochastic searching property and operating on the parameter object directly. With the advantage of this algorithm, the complexity of computing parameters directly can be avoided. In this paper, the genetic algorithm and template matching is integrated. The method was tested using a remote sensing image from Beijing Olympic project area. Results showed that this method has high speed owing to the introduction of Genetic Algorithm into template matching and its practical operation is very simple.
出处
《影像技术》
CAS
2006年第3期44-48,共5页
Image Technology
关键词
遥感影像
目标定位
模板匹配
遗传算法
Remote sensing image
Object Localization
Template Matching
Genetic Algorithm
