摘要
在南方水稻遥感监测中,单一传感器影像数据已不能满足监测精度的要求,需要将高空间分辨率全色影像与中高空间分辨率多光谱影像进行融合,得到新的高空间分辨率多光谱影像,有利于改善影像识别与分类精度.该文利用江苏省金湖地区HJ-1A卫星30m分辨率多波段影像与ALOS卫星2.5m分辨率全色影像进行水稻监测,采用4种融合方法(Brovey变换、IHS变换、高通滤波和小波变换)对2种影像进行融合处理.随后对各种融合影像结果进行了目视定性和融合评价指标定量说明与评价,结果表明小波变换在空间与光谱信息上具有最佳的融合效果.进一步利用小波变换的融合影像进行水稻识别与面积提取,统计表明融合影像相比HJ-1A多光谱影像,水稻面积估测精度从79.26%提高到91.65%.因此,利用多源遥感数据融合的方法对南方水稻面积进行监测,可显著提高其监测精度.
In the remote sensing monitoring of rice cultivation area in South China, the single sensor image data can not meet the requirements of monitoring accuracy. High spatial resolution panchromatic images and medium spatial resolution multispectral images need to be fused into a new high spatial resolution multi-spectral image, which would help to improve the accuracy of image recognition and classification. In a study reported in this paper, HJ-1A satellite 30 m resolution multi-band image and ALOS satellite 2.5 m resolution panchromatic image were used to monitor the rice area in Jinhu Region of Jiangsu Province and the two source data were fused with four fusion methods, i.e. Brovey transformation, IHS transforma- tion, high-pass filter and wavelet transformation. Then, based on a series of fusion indexes （mean, standard deviation, average gradient, correlation coefficient, deviation index and cross-entropy）, the results of the four fusion images were evaluated and described. The wavelet transformation method gave the best results for space and spectral information fusion. Afterwards, wavelet transformation was used to extract the rice growing area. Based on ground truth data for accuracy validation, the results showed that the precision of the estimated rice area increased from 79.26% for the HJ-1A image to 94.65% for the fused image. Therefore, the use of multi-source remote sensing data fusion method for monitoring the rice area in South China can significantly improve the monitoring accuracy.
出处
《西南大学学报（自然科学版）》
CAS
CSCD
北大核心
2012年第6期18-24,共7页
Journal of Southwest University(Natural Science Edition)
基金
国家自然科学基金资助项目(41171336)
江苏省农业科技自主创新基金资助项目(CX-11-2043)
江苏省自然科学基金资助项目(BK2011684)
关键词
遥感数据
融合方法
评价
水稻面积监测
remote sensing data
fusion method
evaluation
rice area monitoring
