A Dark and Bright Channel Prior Guided Deep Network for Retinal Image Quality Assessment
This work addresses quality assessment for retinal images, which is crucial for diagnosing retinal diseases, but it is incremental as it builds on existing deep models by incorporating specific priors.
The paper tackles retinal image quality assessment by proposing GuidedNet, a deep network that embeds dark and bright channel priors into the start layer to improve feature discrimination, achieving effectiveness validated on public and re-annotated datasets.
Retinal image quality assessment is an essential task in the diagnosis of retinal diseases. Recently, there are emerging deep models to grade quality of retinal images. Current state-of-the-arts either directly transfer classification networks originally designed for natural images to quality classification of retinal images or introduce extra image quality priors via multiple CNN branches or independent CNNs. This paper proposes a dark and bright channel prior guided deep network for retinal image quality assessment called GuidedNet. Specifically, the dark and bright channel priors are embedded into the start layer of network to improve the discriminate ability of deep features. In addition, we re-annotate a new retinal image quality dataset called RIQA-RFMiD for further validation. Experimental results on a public retinal image quality dataset Eye-Quality and our re-annotated dataset RIQA-RFMiD demonstrate the effectiveness of the proposed GuidedNet.