Junlin Chen

Learning from Model Rankings Improves Blind Super-Resolution Image Quality Assessment
IEEE International Conference on Systems, Man, and Cybernetics (SMC), 2025
Junlin Chen, Peibei Cao, Guangtao Zhai, Xiaokang Yang, Weixia Zhang*

We propose a data-efficient blind image quality assessment (BIQA) approach for real-world super-resolution (SR). By leveraging global rankings of SR algorithms derived from human perception, our method enhances SR-IQA models without the need for explicit human labels.

[Code] · [Paper]

DCP-GAN: A Novel Deep Learning Network for Fingerprint Super-Resolution
Digital Image Processing and Machine Vision, 2023
Junlin Chen, Rizhi Ding, Khushal Haider Syed

We propose a generative adversarial network with dense connection pyramids, to enhance low-quality fingerprint images and improve recognition accuracy.

[Code] · [Paper]