Fig. 1. Framework of the proposed method

Fig. 2. Differences between visible and infrared images

Fig. 3. Keypoints acquisition

Fig. 4. Pseudo-twin neural network framework

Fig. 5. Fusion layer

Fig. 6. Data set presentation （The first line pertain to the VEDAI datasets， and the second line pertain to the MTV datasets.）

Fig. 7. Experimental results of PCMM-Net on VEDAI dataset

Fig. 8. Experimental results of PCMM-Net on MTV dataset

Fig. 9. Matching accuracy of different algorithms at various pixel thresholds ε in the VEDAI and MTV Datasets

Fig. 10. MTV image registration results

Fig. 11. Matching results of VEDAI dataset original map and affine transform