多源遥感影像高精度自动配准方法研究

针对遥感影像本身存在纹理差异,特征点提取不均匀的自动配准问题,提出了一种新的多源遥感影像高精度自动配准方法。首先,对Harris算子进行了改进,使其提取特征点覆盖范围更广;然后,对影像进行分块处理,对每块用文章改进的Harris算子进行特征点提取,并采用分块迭代剔除策略,保证了特征点在整幅影像中的均匀分布和非冗余分布;其次提出了金字塔-双向-最小二乘多匹配算法进行影像匹配,完成了最终同名点选取;最后利用小三角形面元TIN对影像进行了精确纠正。实验结果表明本文方法获取同名点均匀合理,能够更好地实现多源遥感影像的高精度配准,具有一定实际意义。

High accuracy-automatic registration method research on multi-source remote sensing image

In remote sensing image automatic registration, there are some problems, for example, there are differences between texture image itself and unevenness in feature points extracting. To this problem, the paper proposes a new high precision automatic registration method of multi-source remote sensing image. First of all, the Harris operator was improved to extract feature points covering a wider range; and then, the image is divided into blocks, using the improved Harris operator to extract feature points for each block, and then use the block iteration to eliminate wrong feature points, ensuring the distribution of feature points is uniformly and non redundant in the whole image. Secondly the paper put forward the pyramid-bidirectional-least squares matching algorithm for image matching, completed selection of the final corresponding points. Finally, the paper made use of the TIN to complete accurately correction. The experimental results show that the proposed method of the paper can ensure the corresponding points uniform and reasonable, and it can achieve high precision registration of multi-source remote sensing image, which has certain practical significance.