A Super-Resolution Reconstruction Method of Pupil and Iris Image Based on Multi-Path Interconnected Neural Network

Abstract

Pupil iris images are of great significance to doctors in ophthalmology diagnosis. Generally Due to hardware limitations, it is unable to get a clearer pupil iris images, which is not conducive to the diagnosis process. Aiming at the problem that most network structures This paper proposes a multi - pass - connected remote sensing image super-resolution reconstruction method that can not sufficiently utilize hierarchical features in pupil iris reconstruction. Which can achieve a good reconstruction effect on pupil iris images. First, use the feature extraction network to obtain the low frequency function of the image, and use it as the input of the two sub-networks. Secondly, use multi-path connection to connect different feature fusion units improve connection between fusion units, extract more effective features, and improve the utilization of hierarchical features. Finally, after fusing the features obtained by the two sub-networks, the residual learning always complete the reconstruction of the high-resolution image. Through experimental verification, it is concluded that the reconstructed pupil and iris image of the method suggested in this paper has improved various evaluation indexes compared with the traditional method.