CAJ | 학술논문

针对目前图像拼接算法存在对于图像配准过程中对应特征点对难以准确匹配的问题，提出了一个通过改进的 SURF 算法提取图像特征点，然后对得到的特征点进行描述，利用快速RANSAC 算法配准图像，最后采用像素加权的方法进行图像融合。实验结果表明，提出的改进 SURF方法有效地提高了特征点提取的准确性，去除了错误的匹配点对，将整个拼接过程的效率从之前的13.03对/秒提升到15.20对/秒。
침대목전도상병접산법존재대우도상배준과정중대응특정점대난이준학필배적문제，제출료일개통과개진적 SURF 산법제취도상특정점，연후대득도적특정점진행묘술，이용쾌속RANSAC 산법배준도상，최후채용상소가권적방법진행도상융합。실험결과표명，제출적개진 SURF방법유효지제고료특정점제취적준학성，거제료착오적필배점대，장정개병접과정적효솔종지전적13.03대/초제승도15.20대/초。
Image stitching is mainly used in aerial image processing, medical image analysis, virtual reality technology, computer vision, etc. For image registration, the difficult is to accurately extract the corresponding feature points. This paper puts forward a kind of algorithm based on improved SURF algorithm to extract image feature points, and then describes the feature points, using the RANSAC algorithm to registration the image. Finally, it uses the method of weighted pixel in image fusion. The experiment results show that this method improved the accuracy of feature extract, wipe out the wrong matching points, and improved the processing of stitching from 13.03/s to 15.20/s.