SALIENCY MAP RETRIEVAL FOR ARTISTIC PAINTINGS INSPIRED FROM HUMAN UNDERSTANDING

摘要

This paper presents a simple and efficient method for detecting salient regions in digital representations of paintings. The main challenge is to model the way human eye and mind see and understand visual art. Based on a combination of features such as shape, colour, local contrast and position, the most relevant areas of a digital representation of a painting are detected. The model follows thoroughly the human interpretation of the artistic painting. The presented approach shows robustness regardless of the art movement the analyzed painting belongs to.