Object-Based Image Analysis of Hyper Spectral Imagery Using Semantic Segmentation Techniques

Abstract

Object-based image analysis of Hyperspectral Imagery using Semantic Segmentation strategies is a singular approach for analyzing far-off sensing statistics. This method leverages the energy of a superior system gaining knowledge of (ML) and computer vision algorithms to analyze multidimensional hyperspectral image datasets. The goal is to robustly organize pixels into clusters in step with their spectral and spatial traits. Those clusters are then used to give meaningful records approximately the content material of the photo., the enter pix are pre-processed to reduce noise and boom contrast. A semantic segmentation algorithm is then used to generate excessive-degree masks of the items of interest. The outcomes of those masks are mixed with the input hyperspectral statistics to create several feature vectors describing the spectral and texture homes of every cluster. Sooner or later, a device-mastering algorithm categorizes the gadgets consistent with their traits, presenting precise information about the items in the picture.