Blind Image Quality Assessment Based on Classification Guidance and Feature Aggregation

• DOI: 10.3390/electronics9111811
• Publication Date: 2020-11-02T14:04:46Z
• Authors: Weipeng Cai; Cien Fan; Lian Zou; Yifeng Liu; Yang Ma; Minyuan Wu
• Keywords: Feature (linguistics); Distortion (music); Representation; Contextual image classification

In this work, we present a convolutional neural network (CNN) named CGFA-CNN for blind image quality assessment (BIQA). A unique two-stage strategy is utilized which firstly identifies the distortion type in an using Sub-Network I and then quantifies II. Different from most deep networks, extract hierarchical features as descriptors to enhance representation design feature aggregation layer end-to-end training manner applying Fisher encoding visual vocabularies modeled by Gaussian mixture models (GMMs). Considering authentic distortions synthetic distortions, contains characteristics of CNN trained on self-built dataset ImageNet. We evaluated our algorithm four publicly available databases, results demonstrate that has superior performance over other methods both databases.