Figure and Landscape Photo Quality Assessment Based on Visual Aesthetics

摘要

Among several traditional photo quality assessments, there is yet no one for aesthetic evaluation of figure and landscape photos. So this paper proposes an innovative algorithm to solve this problem. Firstly, according to the theoretical photography rules, we can extract the aesthetic features and color harmonization features from the photo to assess its composition deviation. Then we learn the mapping models between the ground real aesthetics and the aesthetic features. At last, we can use these models to rate the aesthetics of photos. The experiments validate that the proposed algorithm has a better assessment precision and accuracy, which may be widely used in automatic aesthetic appraisal field.