CAJ | 학술논문

针对复杂地面目标前视红外图像边缘模糊、背景杂乱无章,无直接可用的,同源基准图的情况,研究了一种灰度与形状特征相结合的由粗到精景象匹配算法.在粗匹配阶段,将积分图像引入归一化积相关算法中,并对其计算公式进行了近似,使运算速度大大加快;精匹配阶段,利用实时图与所制备基准图形状相似的特点,采用形状上下文进行匹配.实验结果显示,该算法匹配精度高、速度快,对于复杂地面目标前视红外图像的匹配识别具有一定的应用价值.
침대복잡지면목표전시홍외도상변연모호、배경잡란무장,무직접가용적,동원기준도적정황,연구료일충회도여형상특정상결합적유조도정경상필배산법.재조필배계단,장적분도상인입귀일화적상관산법중,병대기계산공식진행료근사,사운산속도대대가쾌;정필배계단,이용실시도여소제비기준도형상상사적특점,채용형상상하문진행필배.실험결과현시,해산법필배정도고、속도쾌,대우복잡지면목표전시홍외도상적필배식별구유일정적응용개치.
According to the FLIR image of complex ground target have blurred edge and background clutter,and no base image available,this paper studied an algorithm that combined the gray information with shape feature to achieve precision gui-dance.In course of coarse matching introduced the integral image into NPROD algorithm and approximated its formula, the efficiency of algorithm had been improved obviously.In the process of fine matching,based on characteristic of the real image and base image had similar shape, the algorithm adopted the shape context descriptor.The experimental results show that the method has highly precision and quickly performance. The algorithm has highly practicality for complex ground garget.