SEMI-AUTOMATED FEATURE EXTRACTION FOR ROOFTOP SOLAR POTENTIAL ASSESSMENT

摘要

The problem of assessing the rooftop solar potential of large habitat is a big challenge for researchers. To address this problem, an approach has been devised to semi-automate the process of extracting the features. The objective of this research is to assess the solar potential of the rooftop features extracted from the high-resolution satellite image at the specific area. An approach using Matlab programming has been devised to extract the features from the high-resolution satellite image using the k-means clustering algorithm. The high-resolution multispectral satellite image of an area within the Haridwar district has been used in this stud). The satellite image has been downloaded from the Google Earth. The k-means clustering algorithm mentioned above has been implemented in Matlab to extract rooftops of the selected study area of Haridwar. India. Accuracy assessment has been performed using rooftop's area obtained by digitization in QGIS. This methodology helped in extracting almost all the rooftops available in the satellite image. The area estimation from the automated rooftops extracted and digitization is matching by 86%. This approach helped in lowering the time required in predicting the solar potential of the densely populated large areas for installing solar photovoltaic modules. This approach will also help in predicting the solar potential of the large fields, barren land, water bodies, etc. in a fast and accurate manner.