Table 10 Performance comparison of models in complex scenarios.

From: A novel approach to palm vein image segmentation combining multi-scale convolution and swin-transformer networks

Model name

PolyU-abrupt lighting change (PSNR = 22 dB)

CASIA-Gaussian noise (\(\sigma =0.1\))

CASIA-rotation deformation (\(30^\circ\))

Comprehensive robustness index

U-Net

88.3/0.812

79.5/0.753

75.2/0.698

0.741

ResNet

89.5/0.826

82.7/0.781

78.9/0.721

0.776

Swin-transformer

91.2/0.843

85.6/0.802

83.4/0.774

0.806

Our model

96.7/0.901

91.3/0.872

92.8/0.865

0.879

Back to article page