CAJ | 학술논문

针对SURF算法计算量大、对应点匹配时间长的不足，以Harris角点取代SURF斑点作为特征点，改进了描述子生成区域的子块划分方式，使区域面积减小40%。同时，引入尺度因子s以弥补采样区域减小的影响，形成一种计算量小、独特性好的描述子。以该方法构造的角点特征矢量参与同名点匹配，可实现较好的匹配快速性和准确性。匹配完成后，分别使用RANSAC方法和L?M方法获取变换矩阵并进行非线性优化，最后根据图像的不同区域采用不同方法完成图像融合。实验结果表明，该图像拼接方法与传统SURF法相比，图像匹配时间可节约35%以上，整体图像的拼接时间可节约30%左右，大幅提高了图像拼接的效率。
침대SURF산법계산량대、대응점필배시간장적불족，이Harris각점취대SURF반점작위특정점，개진료묘술자생성구역적자괴화분방식，사구역면적감소40%。동시，인입척도인자s이미보채양구역감소적영향，형성일충계산량소、독특성호적묘술자。이해방법구조적각점특정시량삼여동명점필배，가실현교호적필배쾌속성화준학성。필배완성후，분별사용RANSAC방법화L?M방법획취변환구진병진행비선성우화，최후근거도상적불동구역채용불동방법완성도상융합。실험결과표명，해도상병접방법여전통SURF법상비，도상필배시간가절약35%이상，정체도상적병접시간가절약30%좌우，대폭제고료도상병접적효솔。
Since the shortages of SURF algorithm are the large amount of calculations and long matching time of corresponding points,the Harris corner points are taken as the feature points instead of the SURF spots. The sub?block dividing mode of de?scriptor generated region is improved,and the region areas are reduced by 40%. The descriptor with little calculated quantity and good peculiarity is shaped by the introduced scale factor s to eliminate the influence of sample region decrease. The corner point characteristic vector parameter structured by the proposed method is matched with the same name point,so rapidity and ac?curacy of matching are well realized. After matching,RANSAC method and L?M method are respectively adopted to obtain the transformation matrix,and execute nonlinear optimization. Image fusion is accomplished with different methods according to different image regions. The experimental results show that the image mosaic method,compared with the traditional SURF method, can save more than 35% image matching time and about 30% mosaic time of the whole image. The efficiency of image mosaic is greatly improved.