论文部分内容阅读
针对以往多源影像中人工选取同名点费时费力并且精度低的问题,本文首先利用尺度不变特征变换(SIFT)特征匹配得到的同名像点作为初值进行仿射变换,生成变换后影像,然后对基准影像与变换后影像进行相关系数匹配,并利用双向匹配删除误匹配点,最后通过仿射逆变换得到原始影像上的同名像点位置。利用此方法对资源三号影像后视影像和实践九号全色影像进行试验,试验证明其结果明显优于在原始影像上直接进行相关系数匹配,可以很好地代替人工选点工作,大大减小了人工选取同名点的工作量。
Aiming at the problem of time-consuming and low-accuracy in the artificial selection of the same name in multi-source image, the paper firstly affine transformed the same-name point obtained by SIFT feature matching as initial value to generate transformed image, and then Matching the correlation coefficient between the reference image and the transformed image, and deleting the matching point by bidirectional matching. Finally, the position of the same name point on the original image is obtained by the affine inverse transformation. Using this method, we test the resource 3 image and the practice 9 panchromatic image, and the experiment proves that the result is obviously better than directly matching the correlation coefficient in the original image, which can well replace the artificial point selection and greatly reduce Small manual selection of the same name of the workload.