Abstract
Most of the traditional supervised classification methods using full-polarimetric synthetic aperture radar (PolSAR) imagery are dependent on sufficient training samples, whereas the results of pixel-based supervised classification methods show a high false alarm rate due to the influence of speckle noise. In this paper, to solve these problems, an object-based supervised classification method with an active learning (AL) method and random forest (RF) classifier is presented, which can enhance the classification performance for PolSAR imagery. The first step of the proposed method is used to reduce the influence of speckle noise through the generalized statistical region merging (GSRM) algorithm. A reliable training set is then selected from the different polarimetric features of the PolSAR imagery by the AL method. Finally, the RF classifier is applied to identify the different types of land cover in the three PolSAR images acquired by different sensors. The experimental results demonstrate that the proposed method can not only better suppress the influence of speckle noise, but can also significantly improve the overall accuracy and Kappa coefficient of the classification results, when compared with the traditional supervised classification methods.
Highlights
Land-cover classification of remote sensing imagery is becoming more and more important for local and regional planning [1,2], environmental impact assessment [3,4], agriculture monitoring [5], etc
The generalized statistical region merging (GSRM) method is used to reduce the impact of inherent speckle noise; the active learning (AL) algorithm is applied to select reliable training samples from the different polarimetric features of the polarimetric synthetic aperture radar (PolSAR) imagery; and the random forest (RF) classifier is used as the classifier to identify the different types of land cover in the PolSAR images
We used the AL algorithm to select an effective training set with high representation quality and low redundancy; the RF classifier was used as the classifier to identify the different types of land cover in the AIRSAR images; and, we evaluated the classification accuracy of the different methods
Summary
Land-cover classification of remote sensing imagery is becoming more and more important for local and regional planning [1,2], environmental impact assessment [3,4], agriculture monitoring [5], etc. PolSAR images are subject to the influence of inherent speckle noise and still have finely divided spots despite speckle filtering, so the results of the traditional pixel-based supervised classification methods show a high false alarm rate. Supervised classification using full-polarimetric SAR data remains a challenge In this regard, it is necessary to design a supervised classification method for PolSAR imagery that uses as few labeled samples as possible to obtain better precision. The GSRM method is used to reduce the impact of inherent speckle noise; the AL algorithm is applied to select reliable training samples from the different polarimetric features of the PolSAR imagery; and the RF classifier is used as the classifier to identify the different types of land cover in the PolSAR images.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have