基于SURF特征的侧扫声呐图像精确镶嵌

Accurate Mosaic of Side Scan Sonar Images Based on SURF Features

针对侧扫声呐条带拼接过程中容易出现错位、灰度不均衡等问题,本文研究了一种基于SURF算法的侧扫声呐图像精确镶嵌方法。该方法通过USM滤波对单幅条带图像进行特征增强,然后利用SURF算法对相邻条带图像进行镶嵌处理,SURF算法具有执行速度快、信息量小、匹配精度高、提取的特征不受光照变化、透视、仿射、旋转等影响等优点。针对条带之前存在的误匹配点,采用RANSAC算法进行自动剔除,最后采用小波融合法对图像进行精确镶嵌。实验研究证明,采用本方法能够有效解决匹配过程中目标错位等问题,实现了侧扫声呐图像的精确镶嵌。

In view of the problems such as dislocation and uneven gray scale in the splicing process of side-scan sonar strip,this paper studies an accurate mosaic method of side-scan sonar image based on the SURF algorithm.In this method,USM filter is used to highlight the features of a single strip image,and the SURF algorithm is applied to mosaic adjacent strip images.The SURF algorithm has the advantages of fast execution,small amount of information,high matching accuracy,and the extracted features are not affected by illumination changes,perspective,affine,rotation and many other factors.The RANSAC algorithm is adopted to eliminate the false matching points before the strip,and finally wavelet fusion is used to mosaic the image accurately.The experiment shows that the method proposed in this paper can effectively solve the problem of target misalignment in the matching process and achieve precise mosaic of side-scan sonar image.