Texture-Guided Multiscale Feature Learning Network for Palmprint Image Quality Assessment

Xiao Sun, Lunke Fei, Shuping Zhao, Shuyi Li, Jie Wen, Wei Jia

Abstract: Palmprint recognition has attracted widespread attention because of its advantages such as easy acquisition, rich texture, and security. However, most existing palmprint recognition methods focus most on feature extraction and matching without evaluating the quality of palmprint images, possibly leading to low recognition efficiency. In this paper, we propose a texture-guided multiscale feature learning network for palmprint image quality assessment. Specifically, we first employ a multiscale feature learning network to learn multiscale features. Then, we simultaneously use the multiscale features to learn image quality features by a QualityNet and texture features by a texture guided network. Texture features are then further used to learn texture quality features via TextureNet. Finally, we fuse the image quality features and texture quality features as palmprint quality features to predict the quality score via a regressor. Experimental results on the widely used palmprint database demonstrate that the proposed method consistently outperforms the state-of-the-art methods on palmprint image quality assessment.