基于SURF和矩阵乘法的超大规模遥感图像亚像素配准算法研究

发布时间：2018-08-13 16:35

【摘要】：图像配准是一项基本而又非常关键的图像预处理技术。在很多应用领域,要求配准精度达到亚像素级。现有的相位相关法具有精度高、计算简单等特点,但是随着图像规模的增大,计算付出的时间代价是巨大的。本文提出基于SURF和矩阵乘法相位相关法的超大规模遥感图像亚像素配准算法,采用化整为零的方法,首先把整幅图像划分成不同区域,其次使用改进的Canny算法进行边缘分割,去除无用信息,再次使用SURF算法提取特征,最后在关键点周围使用矩阵乘法相位相关估计图像亚像素偏移量。实验表明本文提出的算法不仅提高了算法运行速度,同时也解决了图像尺寸太大导致一般计算机无法处理的问题。并且由于矩阵乘法相位相关的良好抗噪声特性,因此即使存在噪声,算法仍然可以获得较高的亚像素偏移量估计精度。
[Abstract]:Image registration is a basic and very important image preprocessing technique. In many applications, registration accuracy is required to reach sub-pixel level. The existing phase correlation methods have the advantages of high accuracy and simple calculation, but with the increase of the image size, the time cost of the calculation is enormous. In this paper, a sub-pixel registration algorithm for super-large scale remote sensing images based on SURF and matrix multiplication phase correlation method is proposed. Firstly, the whole image is divided into different regions, and then the improved Canny algorithm is used for edge segmentation. SURF algorithm is used to extract the feature again and matrix multiplication phase correlation is used to estimate the sub-pixel offset around the key points. Experimental results show that the proposed algorithm not only improves the speed of the algorithm, but also solves the problem that the image size is too large to be processed by a computer. Because of the good anti-noise characteristic of matrix multiplication phase correlation, even if there is noise, the estimation accuracy of sub-pixel offset can be obtained.
【作者单位】：中国科学院上海技术物理研究所;
【基金】：全球变化与海汽相互作用专项(GASI-03-03-01-01)
【分类号】：TP751

【参考文献】