Feature point detection for optical and SAR remote sensing images registration

Table 1. Repetition rates of the four methods in the case of nonlinear radiometric differences
View table
View in Article
Table 1. Repetition rates of the four methods in the case of nonlinear radiometric differences
Group Harris
SAR-
Harris
m+M-Harris MMPC-Harris
Group 1 72.85% 41.76% 64.51% 87.91%
Group 2 60.50% 71.56% 65.13% 90.53%

Table 2. Program logic diagram of MMPC-Harris

View table

View in Article

Table 2. Program logic diagram of MMPC-Harris

MMPC-Harris算法

输入图像

1）分块策略：将图像分成 $S_{n} \times S_{m}$ 个子图像；

2）分别对每个子图像计算PC值；

for k = 1， $\cdot \cdot \cdot$ ，5 do

｛ $k_{t} = - 1 + 0.5 (k - 1)$ ，构建 $M_{k}$ ，Harris特征点提取；｝

end for

3） for i = 1， $\cdot \cdot \cdot$ ，（角点数） do

统计特征点i在 $M_{1} ~ M_{n}$ 上出现的次数；

if （出现次数）≥3 do

｛计算该特征点在 $M_{1} ~ M_{n}$ 上的坐标的平均值，作为该特征点最终的坐标；｝

end if

end for

4）合并特征点

输出： MMPC-Harris检测结果

Table 2. NCM and RMSE of the four algorithms
View table
View in Article
Table 2. NCM and RMSE of the four algorithms
Algorithm RMSE NCM
Harris+HOSMI 2.82 23
SAR-Harris+HOSMI 2.66 32
m+M-Harris+HOSMI 1.91 35
MMPC-Harris+HOSMI 1.67 58

Tools

Get Citation

Copy Citation Text

Lina WANG, Huaidan LIANG, Zhongshi WANG, Rui XU, Guangfeng SHI. Feature point detection for optical and SAR remote sensing images registration[J]. Optics and Precision Engineering, 2022, 30(14): 1738

Download Citation

EndNote(RIS)BibTex Plain Text

Set citation alerts for article

Save article for my favorites