基于最大公约数的遥感影像空间尺度转换算法
作者简介:高永刚(1976-),男,博士,讲师,主要从事遥感图像处理和卫星测高研究。E-mail: yggao@fzu.edu.cn
收稿日期: 2015-05-11
要求修回日期: 2015-08-03
网络出版日期: 2015-12-20
基金资助
福建省自然科学基金项目(2012J01171、2012J01169)
国家科技支撑计划项目(2013BAC08B01-05)
福建省教育厅科技项目(JA15064)
海岛(礁)测绘技术国家测绘地理信息局重点实验室基金项目(2010B09)
Scale Transformation Algorithm for Remote Sensing Imagery Based on Greatest Common Divisor
Received date: 2015-05-11
Request revised date: 2015-08-03
Online published: 2015-12-20
Copyright
多源、多尺度遥感影像为研究不同尺度的地表变化提供了丰富的数据。但其在作比较研究时,通常会涉及空间尺度统一问题,当多源遥感影像之间的空间分辨率为非整倍数关系时,其空间尺度统一相对困难。为此,本文针对多源、多尺度遥感影像间尺度比较时所涉及的空间尺度转换问题,提出了最大公约数的空间尺度转换算法,并以IKONOS多光谱影像为数据源,采用若干商业软件和本文所提算法进行空间尺度转换比较实验;同时,利用均值、标准差和相关系数等6个评价指标对空间尺度变换后的影像进行定量评价。结果表明,本文提出的空间尺度转换方法对原始影像的光谱信息等特征具有很好的保真性,简单易行,可实现遥感影像任意空间尺度的转换,解决了多源遥感影像之间的空间分辨率为非整倍数关系时的空间尺度转换问题。
高永刚 , 徐涵秋 . 基于最大公约数的遥感影像空间尺度转换算法[J]. 地球信息科学学报, 2015 , 17(12) : 1520 -1528 . DOI: 10.3724/SP.J.1047.2015.01520
The abundant remote sensing data with various spatial, radiational and spectral resolutions from multi-platforms provide rich information sources for the study of land surface information changes at different scales. Scale variation and sensitivity have a great impact on the application of remote sensing imagery in different scientific fields. We proposed a transformation algorithm to unify the scales for comparing data at different scales. The method is a scale transformation algorithm based on the greatest common divisor (STAGCD). Firstly, the greatest common divisor (GCD) between two different spatial scales is calculated. Secondly, according to the GCD, a GCD image will be produced by resampling the original remote sensing image. Finally, the new scale image will be obtained according to certain intervals for row and column to choose data from the GCD image. Several scale transformation algorithms have been employed in the test of the scale unification for an IKONOS image, including STAGCD and some other algorithms from professional software packages, such as ER Mapper, ERDAS, Matlab and so on. The effectiveness of these algorithms has been evaluated based on the information keeping degree compared with the original remote sensing image. A total of six indicators have been used for quantitative evaluation of the scale transformed images. The histogram and probability density function of Gauss based on kernel bandwidth optimization have been used for visual interpretation of the scale transformed images. The results show that the STAGCD image has adequate ability for keeping the information of original image. When scaling-down, STAGCD only increases the image size, but cannot improve the image’s spatial resolution. When scaling-up, STAGCD not only reduces the image size, but also decreases the image resolution. The STAGCD method is simple and can transform remote sensing imagery at different scales. The method provides an effective solution for the scale transformation between images without an integer multiple relationship.
Fig. 1 Flowchart of indirect comparison method图1 间接比较法流程图 |
Fig. 2 Flowchart of scale transformation experiment图2 尺度转换实验流程图 |
Fig. 3 Original and scale transformation imageries of IKONOS (Scale: 2.1 m)图3 IKONOS原始影像与尺度转换后影像(尺度:2.1 m) |
Fig. 4 Comparison diagram of histogram of original and scale transformation imageries (Scale: 2.1 m)图4 IKONOS原始影像与各尺度转换后影像直方图对比图(尺度:2.1 m) |
Tab. 1 Quantitative evaluation of the scale transformation algorithms (Scale: 2.1 m)表1 IKONOS原始影像与尺度转换后影像定量评价表(尺度: 2.1 m) |
方法 | 波段 | 评价指标 | |||||
---|---|---|---|---|---|---|---|
Mean | STD | SD | RMSE | ERGAS | CC | ||
原始影像 | 1 | 457.501 | 88.939 | - | - | - | - |
2 | 499.650 | 134.548 | - | - | - | - | |
3 | 378.310 | 153.395 | - | - | - | - | |
4 | 504.025 | 174.153 | - | - | - | - | |
商业(软件)算法 | 1 | 457.658 | 89.127 | 25.827 | 52.062 | 11.376 | 0.859 |
2 | 499.956 | 134.772 | 40.693 | 77.405 | 15.482 | 0.835 | |
3 | 378.582 | 153.520 | 46.246 | 85.474 | 22.577 | 0.845 | |
4 | 505.493 | 174.515 | 68.193 | 116.083 | 22.964 | 0.779 | |
本文算法 | 1 | 457.501 | 88.939 | 0.000 | 0.000 | 0.000 | 1.000 |
2 | 499.650 | 134.548 | 0.000 | 0.000 | 0.000 | 1.000 | |
3 | 378.310 | 153.395 | 0.000 | 0.000 | 0.000 | 1.000 | |
4 | 504.025 | 174.153 | 0.000 | 0.000 | 0.000 | 1.000 | |
商业(软件)算法→本文算法 | 1 | 457.324 | 88.954 | 4.622 | 19.211 | 4.201 | 0.977 |
2 | 499.254 | 134.478 | 7.528 | 30.255 | 6.060 | 0.974 | |
3 | 377.748 | 153.117 | 8.740 | 35.321 | 9.350 | 0.973 | |
4 | 503.760 | 174.071 | 12.328 | 49.902 | 9.906 | 0.959 | |
本文算法→商业(软件)算法 | 1 | 457.532 | 89.015 | 17.269 | 40.397 | 8.829 | 0.896 |
2 | 499.827 | 134.672 | 27.521 | 61.326 | 12.272 | 0.895 | |
3 | 378.413 | 153.496 | 31.270 | 67.517 | 17.837 | 0.903 | |
4 | 505.196 | 174.238 | 47.259 | 91.853 | 18.218 | 0.860 |
Tab. 2 Quantitative evaluation of the scale transformation algorithms (Scale: 0.6 m)表2 IKONOS原始影像与尺度转换后影像定量评价表(尺度: 0.6 m) |
方法 | 波段 | 评价指标 | |||||
---|---|---|---|---|---|---|---|
Mean | STD | SD | RMSE | ERGAS | CC | ||
原始影像 | 1 | 457.501 | 88.939 | - | - | - | - |
2 | 499.650 | 134.548 | - | - | - | - | |
3 | 378.310 | 153.395 | - | - | - | - | |
4 | 504.025 | 174.153 | - | - | - | - | |
商业(软件)算法 | 1 | 457.627 | 89.206 | 18.418 | 42.521 | 9.291 | 0.893 |
2 | 499.810 | 134.880 | 29.469 | 64.073 | 12.820 | 0.886 | |
3 | 378.526 | 153.714 | 33.565 | 71.888 | 18.991 | 0.890 | |
4 | 504.407 | 174.265 | 49.750 | 100.117 | 19.848 | 0.834 | |
本文算法 | 1 | 457.501 | 88.939 | 0.000 | 0.000 | 0.000 | 1.000 |
2 | 499.650 | 134.548 | 0.000 | 0.000 | 0.000 | 1.000 | |
3 | 378.310 | 153.395 | 0.000 | 0.000 | 0.000 | 1.000 | |
4 | 504.025 | 174.153 | 0.000 | 0.000 | 0.000 | 1.000 | |
商业(软件)算法→本文算法 | 1 | 457.683 | 89.222 | 15.082 | 35.732 | 7.807 | 0.925 |
2 | 499.734 | 134.973 | 23.810 | 53.983 | 10.803 | 0.920 | |
3 | 378.419 | 153.820 | 27.105 | 59.563 | 15.740 | 0.916 | |
4 | 504.108 | 174.187 | 39.576 | 79.526 | 15.776 | 0.896 | |
本文算法→ 商业(软件)算法 | 1 | 457.600 | 89.072 | 17.133 | 40.822 | 8.921 | 0.899 |
2 | 499.773 | 134.750 | 27.604 | 61.405 | 12.286 | 0.898 | |
3 | 378.509 | 153.654 | 31.560 | 69.278 | 18.303 | 0.897 | |
4 | 504.139 | 174.248 | 46.784 | 97.203 | 19.281 | 0.844 |
Tab. 3 Quantitative evaluation of the scale transformation algorithms (Scale: 1.5 m)表3 IKONOS原始影像与尺度转换后影像定量评价表(尺度: 1.5 m) |
方法 | 波段 | 评价指标 | |||||
---|---|---|---|---|---|---|---|
Mean | STD | SD | RMSE | ERGAS | CC | ||
原始影像 | 1 | 457.501 | 88.939 | - | - | - | - |
2 | 499.650 | 134.548 | - | - | - | - | |
3 | 378.310 | 153.395 | - | - | - | - | |
4 | 504.025 | 174.153 | - | - | - | - | |
商业(软件)算法 | 1 | 457.537 | 88.939 | 19.808 | 43.908 | 9.597 | 0.893 |
2 | 499.698 | 134.478 | 31.727 | 66.310 | 13.270 | 0.878 | |
3 | 378.434 | 153.370 | 36.363 | 74.707 | 19.741 | 0.882 | |
4 | 504.384 | 174.331 | 53.269 | 103.042 | 20.429 | 0.826 | |
本文算法 | 1 | 457.501 | 88.939 | 0.000 | 0.000 | 0.000 | 1.000 |
2 | 499.650 | 134.548 | 0.000 | 0.000 | 0.000 | 1.000 | |
3 | 378.310 | 153.395 | 0.000 | 0.000 | 0.000 | 1.000 | |
4 | 504.025 | 174.153 | 0.000 | 0.000 | 0.000 | 1.000 | |
商业(软件)算法→本文算法 | 1 | 457.370 | 88.746 | 17.231 | 40.888 | 8.940 | 0.898 |
2 | 499.360 | 134.352 | 27.665 | 61.670 | 12.350 | 0.895 | |
3 | 378.085 | 153.354 | 31.828 | 69.785 | 18.458 | 0.890 | |
4 | 503.460 | 174.258 | 46.045 | 95.918 | 19.052 | 0.849 | |
本文算法→商业(软件)算法 | 1 | 457.456 | 88.861 | 18.961 | 42.756 | 9.348 | 0.901 |
2 | 499.597 | 134.493 | 30.226 | 64.901 | 12.996 | 0.882 | |
3 | 378.419 | 153.359 | 34.431 | 71.901 | 19.015 | 0.889 | |
4 | 504.299 | 174.313 | 52.045 | 97.358 | 19.338 | 0.843 |
Tab. 4 Quantitative evaluation of the scale transformation algorithms (Scale: 3 m)表4 IKONOS原始影像与尺度转换后影像定量评价表(尺度: 3 m) |
方法 | 波段 | 评价指标 | |||||
---|---|---|---|---|---|---|---|
Mean | STD | SD | RMSE | ERGAS | CC | ||
原始影像 | 1 | 457.501 | 88.939 | - | - | - | - |
2 | 499.650 | 134.548 | - | - | - | - | |
3 | 378.310 | 153.395 | - | - | - | - | |
4 | 504.025 | 174.153 | - | - | - | - | |
商业(软件)算法 | 1 | 457.370 | 88.746 | 17.311 | 40.973 | 8.955 | 0.899 |
2 | 499.360 | 134.352 | 27.686 | 61.452 | 12.297 | 0.896 | |
3 | 378.085 | 153.354 | 31.668 | 69.343 | 18.320 | 0.891 | |
4 | 503.460 | 174.258 | 46.766 | 97.201 | 19.279 | 0.845 | |
本文算法 | 1 | 457.501 | 88.939 | 0.000 | 0.000 | 0.000 | 1.000 |
2 | 499.650 | 134.548 | 0.000 | 0.000 | 0.000 | 1.000 | |
3 | 378.310 | 153.395 | 0.000 | 0.000 | 0.000 | 1.000 | |
4 | 504.025 | 174.153 | 0.000 | 0.000 | 0.000 | 1.000 | |
商业(软件)算法→本文算法 | 1 | 457.468 | 89.035 | 11.533 | 34.388 | 7.517 | 0.935 |
2 | 499.554 | 134.604 | 18.368 | 51.286 | 10.266 | 0.929 | |
3 | 378.162 | 153.450 | 21.182 | 58.116 | 15.368 | 0.910 | |
4 | 504.145 | 174.560 | 29.918 | 77.752 | 15.423 | 0.901 | |
本文算法→商业(软件)算法 | 1 | 457.392 | 88.787 | 17.091 | 40.755 | 8.910 | 0.900 |
2 | 499.400 | 134.412 | 27.427 | 61.422 | 12.299 | 0.896 | |
3 | 378.131 | 153.421 | 31.565 | 69.502 | 18.381 | 0.897 | |
4 | 503.449 | 174.265 | 45.571 | 95.478 | 18.965 | 0.850 |
The authors have declared that no competing interests exist.
[1] |
|
[2] |
|
[3] |
|
[4] |
|
[5] |
|
[6] |
|
[7] |
|
[8] |
|
[9] |
|
[10] |
|
[11] |
|
[12] |
|
[13] |
|
[14] |
|
[15] |
|
[16] |
|
[17] |
|
[18] |
|
[19] |
|
/
〈 | 〉 |