Adaptive polar transform and fusion for human face image processing and evaluation

Table 3 Experimental results on UGC-JU face database

Sl. no.	Method		Sensitivity	Specificity	Accuracy
1	Unregistered + No Fusion + PCA + MLP	Visual images	81.91	60.00	77.53
1	Unregistered + No Fusion + PCA + MLP	Thermal images	76.32	54.87	72.06
2	Unregistered + Pixel Fusion + PCA + MLP		86.32	73.53	83.76
3	Unregistered + Wavelet Fusion + PCA + MLP		87.21	77.06	85.18
4	Log Polar Registered + No fusion + PCA + MLP	Visual images	85.44	72.35	82.82
4	Log Polar Registered + No fusion + PCA + MLP	Thermal images	82.79	64.12	79.06
5	Adaptive Log Polar Registered + No fusion + PCA + MLP	Visual images	86.62	75.88	84.47
5	Adaptive Log Polar Registered + No fusion + PCA + MLP	Thermal images	84.26	68.82	81.18
6	Log Polar Registered + Pixel Fusion + PCA + MLP		88.38	80.00	86.71
7	Adaptive Log Polar Registered + Pixel Fusion + PCA + MLP		92.06	82.94	90.24
8	Log Polar Registered + Wavelet Fusion + PCA + MLP		92.94	87.65	91.88
9	Adaptive Log Polar Registered + Wavelet Fusion + PCA + MLP		96.77	91.77	95.77