In this paper, we proposed a complete algorithm of
detecting and segmenting building in high-resolution
RSIs. This algorithm mainly contains three stages: 1)
shadow detection, 2) building detection and rough
segmentation, 3) accurate segmentation based on level set
In the first stage, we combine SLIC, LDA and SVM to
detecting shadow areas; in the second stage, a constrained
regional growth algorithm is presented to find roughly the
locations and areas of the buildings; finally, the accurate
segmentation of buildings is realized based on level set
model. Experimental results showed that the proposed
algorithm is effective, robust and precise than some
competing models.
In future, we will focus on multicategory RSI classification and data mining.
In this paper, we proposed a complete algorithm of detecting and segmenting building in high-resolution RSIs. This algorithm mainly contains three stages: 1) shadow detection, 2) building detection and rough segmentation, 3) accurate segmentation based on level setIn the first stage, we combine SLIC, LDA and SVM to detecting shadow areas; in the second stage, a constrainedregional growth algorithm is presented to find roughly thelocations and areas of the buildings; finally, the accurate segmentation of buildings is realized based on level set model. Experimental results showed that the proposed algorithm is effective, robust and precise than some competing models. In future, we will focus on multicategory RSI classification and data mining.
การแปล กรุณารอสักครู่..
