A Pixel Texture Index Algorithm and Its Application

Abstract

Image segmentation is essential for object-oriented analysis, and classification is a critical parameter influencing analysis accuracy. However, image classification and segmentation based on spectral features are easily perturbed by the high-frequency information of a high spatial resolution remotely sensed (HSRRS) image, degrading its classification and segmentation quality. This article first presents a pixel texture index (PTI) by describing the texture and edge in a local area surrounding a pixel. Indeed.. The experimental results highlight that the HSRRS image classification and segmentation quality can be effectively improved by combining it with the PTI image. Indeed, the overall accuracy improved from 7% to 14%, and the kappa can be increased from 11% to 24%, respectively.