航空滤光片阵列多光谱图像曲面拟合双阈值配准

Dual-threshold registration based on surface fitting of aerial filter array multispectral images

图像配准过程中,匹配点位置精度是决定图像配准精度的关键。本文针对航空滤光片阵列多光谱图像因各谱段间存在像点位移而使误匹配点剔除比较困难的问题,提出了一种基于匹配点位置差曲面拟合双阈值剔除方法。首先,选取多光谱中间波段图像作为基准图像,利用SIFT算法分别提取基准图像和待配准图像的匹配点;其次,在基准图像匹配点处逐点计算两波段图像匹配点的位置差,构建匹配点Delaunay三角网,利用IDW(反距离加权)算法拟合整幅图像位置差曲面;然后,对位置差曲面进行平滑处理,再分别向上向下平移一定容差范围,构成位置差三维阈值空间;最后,利用位置差三维阈值空间筛选出精确匹配点,并完成图像配准工作。理论分析与实验结果表明:该方法可以有效筛选出航空滤光片阵列多光谱图像高精度匹配点,进而有效提高多光谱图像配准的精度。

Aerial filter array multispectral images and their high precision registrations are important for guaranteeing subsequent image processing and application.In the process of image registration,the position accuracy of matching points is important in determining the accuracy of image registration.However,objects of different strips in the same band image are acquired at different moments,the image displacement between single-band images is large,and the difference in geometric errors of matching points between topographic undulating areas and flat areas in the image is obvious.Additionally,false matching points cannot be accurately eliminated by the global matrix.The difficulty of eliminating mismatched points in multispectral images of aerial filter arrays must be addressed because of the displacement of image points between spectral segments.Thus,a new method of double-threshold elimination based on matching point position difference surface fitting is proposed in this study.First,the intermediate band image of the filter array multispectral images was selected as the reference image,and the matching points in the reference image and the image to be registered were extracted by the subpixel-level SIFT algorithm.Second,the difference in the positions of the matching points of the two bands was calculated point by point at the matching points of the benchmark image,and the Delaunay triangulation network of matching points in the reference image was constructed.The position difference surface was smoothed,the position differences between the matching points of the reference image and the corresponding matching points of the image requiring registration were calculated point by point,and a certain tolerance range was shifted upward and downward to form a 3D position difference threshold space.Finally,accurate matching points were selected using the 3D threshold space of the position difference to complete the registration.The three-band composite image of the algorithm-registered image in this study presented clear features and well-defined details and met the requirements of subsequent data processing and application.The effectiveness of the proposed algorithm was illustrated by registering two datasets of filter array multispectral images,from which qualitative and quantitative perspectives were verified.Regarding false color,the composite image processed by the proposed algorithm did not show obvious pseudoedges,and the features were clear.However,pseudoedges were obvious in the comparison algorithm and difference image grayscale histograms.Among the experiments of the two datasets,the difference image histogram curve of the proposed algorithm presented the largest shift to the left.The image registered by the proposed algorithm had the smallest difference from the reference image and the best registration effect.Theoretical analysis and experimental results show that the dual-threshold pointing algorithm based on matching point difference fitting of curved surfaces can screen high-precision matching points in aerial filter array multispectral images and effectively improve the accuracy of image registration.Surface fitting to the position difference of matching points can help reveal the trend of image point displacement in each region.This scheme can also effectively eliminate false matching points around the correct matching points,especially since the image displace.