改进型PSI算法及其在高空间分辨率遥感影像分割中的应用

Improved Pixel Shape Index Algorithm and Its Application in Segmentation of High Spatial Resolution Remote Sensing Imagery

本文针对PSI(Pixel Shape Index,PSI)算法存在的弊端,对其加以改进,并提出了一种改进型PSI算法。相比原算法,新算法的改进主要体现在:1在方向线生成阶段,为了充分考虑不同波段光谱特征的同质性存在的差异,每个波段数据层方向线的生成分开独立进行,这提高了方向线生成的合理性;2提出了每一条方向线长度为各波段数据层的方向线长度的加权和,以进一步体现不同波段光谱特征之间同质性上存在的差异,从而提高像元(尤其是边缘处的像元)PSI值的准确性。最后,通过实验证明:影像分割时,联合改进型PSI派生波段,能有效提高外形规则的人工地物的分割精度,"边缘效应"明显减少。

Aiming at the defects of the Pixel Shape Index (PSI)algorithm,this paper proposed an improved PSI algorithm. The main advantages of the new algorithm are listed as follows:(1 )In order to enhance the rationality of direction line generation,the differences existed in the spectral homogeneity between the bands had been considered adequately.The direction line on each band layer was extended independently.(2 )The length of every direction line was obtained by weighting the lengths of all the direction lines along the same orientation on all band layers.The differences existed in the spectral homogeneity between the bands were further reflected by this way,and the accuracy of PSI about each pixel (especially the edge pixel)was improved.The experimental results showed that the segmentation accuracy of artificial feature with regular shape was improved effectively by combining with the improved PSI derived band during the image segmentation.The “edge effect"phenomenon had been depressed significantly.