Feature Extraction and Classification of Heritage Image from Crowd Source

Abstract

Textures are the most important features that describe an image. To find the relevant and specific information about images and classify them properly, a robust machine learning algorithm should be trained. The paper focuses on identifying monuments by applying statistical texture analysis and classification using SVM algorithm. Our case study is towards the monuments in and around Hubli-Dharwad city that is collected as crowd sourced image database and tested as a part of the Indian digital heritage project