摘要
通过分析电子稳像中灰度投影算法及其相关函数的特点,提出了一种基于单峰特征的改进算法,即三点自适应局域搜索算法:在搜索区间2n+1内以中点为中心,搜索宽度为n选取两个端点和中点共3点,计算并比较其相关值,并以最小值为中心,搜索宽度减半再确定两个端点值,再次比较,直到搜索完毕,找到最佳匹配点的最小值.同全局搜索算法相比,在搜索区间长度为2n+1时,全局算法运算量为2n+1,该算法仅为3n,运算量显著减少.该算法在船载稳像系统中验证表明,该算法的计算量大大降低,较好地满足了稳像系统的实时性要求.
By analyzing the gray projection algorithm and correlation function curves for detecting motion vector in image sequence in electronic image stabilization technique,an improved searching algorithm-three point locally adaptive searching algorithm on the basis of single peak characteristic is proposed,and compared with the full-search method which can cause time delay.The procedures are as follows： regard the middle point as the center,and select three points within the search coverage（2n＋1）： two extremes and the middle point.Calculate and compare their correlation values,then get the min.Then regard the min as the center again and search within the half-reduced coverage so to find the two extremes.The search goes on until the final min is found.Experimental results show its operation load is 3n instead of 2n＋1 of global searching algorithm when searching coverage is 2n＋1.This method has been applied to ship-borne reconnaissance system and confirmed in high calculation speed and high accuracy on the test result.
出处
《微电子学与计算机》
CSCD
北大核心
2008年第11期212-215,共4页
Microelectronics & Computer
关键词
电子稳像
运动矢量
灰度投影算法
自适应搜索算法
electronic image stabilization
motion vector
gray projection algorithm
adaptive search algorithm
作者简介
张永祥 男，（1975-），博士，讲师．研究方向为数字稳像系统、DSP应用、计算机视觉．
