CAJ | 학술논문

图像匹配在计算机视觉和数字摄影测量中都是一项十分重要的任务。为了加速匹配并为高精度子像素匹配提供稳健而相对精确的初值，本文提出了一种基于遗传算法（GA）和最小二乘法匹配（LSM）相结合的遥感图像子像素匹配算法。实验表明：基于遗传算法的匹配算法比序贯相似度度量（SSDA）和经典的平均绝对差（MAD）方法要快得多。以我们的实验数据为例，MAD与GA在相关计算总量上的比至少为36∶1。本文从理论上也证明了这一点。
도상필배재계산궤시각화수자섭영측량중도시일항십분중요적임무。위료가속필배병위고정도자상소필배제공은건이상대정학적초치，본문제출료일충기우유전산법（GA）화최소이승법필배（LSM）상결합적요감도상자상소필배산법。실험표명：기우유전산법적필배산법비서관상사도도량（SSDA）화경전적평균절대차（MAD）방법요쾌득다。이아문적실험수거위례，MAD여GA재상관계산총량상적비지소위36∶1。본문종이론상야증명료저일점。
Image matching is a very important task in either computer visionor digital photogrammetry. In the paper, an approach to remote sensing image matching combining genetic algorithm(GA) with least square matching(LSM) is presented to speed up image matching and provide a robust reliable and relatively accurate initial value for high-precision subpixel matching. The experiment shows that the matching method based on GA is much faster than those based on Sequential Similarity Detection Algorithm(SSDA) and classical Mean Absolute Difference(MAD). Taking our experiment data as an example, the rate of the sum of correlation calculation of MAD to that of GA is at least 36∶1. This fact is proved theoretically.