Image Aesthetics Evaluation Algorithm based on Visual Attention Model
Li Xin Hou, Jinhua Yang · 2023
In recent years, many researchers at home and abroad have paid more and more attention to the in-depth study of the task of image aesthetic quality evaluation, because image aesthetic quality evaluation can play a potential role in many practical applications. Then, this paper proposes the novel image aesthetics evaluation algorithm based on visual attention model. Firstly, the novel 2D OTSU image segmentation algorithm is proposed to conduct the pre-processing taks of the captured images. Then, the novel visual attention model is designed, the proposed model uses the visual association and context double attention mechanism, uses visual association attention to ensure the visual coherence, and uses the context attention to obtain more complete semantic information, and the first step is the image segmentation. Furthermore, the novel evaluation of the image aesthetics is designed and tested. Through the visual model simulation and the ROC testing, the proposed algorithm is proven to be efficient.