Road extraction from very high-resolution (VHR) true color (RGB) images is a popular task in the area of image processing and computer vision. However, there are a number of obstacles that make this task difficult, such as road coverage by tree canopies, shadows, and similar surface objects. To fine-tune the task of road feature extraction, we have proposed a hybrid approach involving various popular image processing functions such as thresholding, fusion, and morphology. Moreover, road extraction from satellite-based VHR-RGB images is quite challenging as compared to doing the same task using multispectral or hyperspectral images. After road extraction, we have also analyzed the efficacy of our proposed approach using several machine learning algorithms. Ultimately, a higher performance with 91.52% accuracy was achieved for the task using the proposed approach.
Chromatic Logical Fusion for Road Feature Extraction in VHR-RGB Satellite Images
23.02.2024
1673701 byte
Aufsatz (Konferenz)
Elektronische Ressource
Englisch
Automatic archaeological feature extraction from satellite VHR images
Online Contents | 2010
|