Fig. 1. Dcomposition process of NSST multi-scale and multi-directional

Fig. 2. Structure of simple PA-PCNN model

Fig. 3. Pixel logarithmic sub-image obtained by two-layer decomposition of low-frequency image. (a) D^(1,0); (b) D^(1,1); (c) D^(2,0); (d) D^(2,1)

Fig. 4. Basic framework of fusion method

Fig. 5. Fusion results under different μ values. (a) μ→0; (b) μ=10; (c) μ=30; (d) μ=50; (e) μ=100; (f) μ=300; (g) μ=500; (h) μ→+¥

Fig. 6. Fusion results of different methods under “Kayak” image. (a) IR image; (b) VIS image; (c) PCA method; (d) LatLRR method; (e) MSD method; (f) PCNN method; (g) PA-PCNN method; (h) proposed method

Fig. 7. Fusion results of different methods under “Road” image. (a) IR image; (b) VIS image; (c) PCA method; (d) LatLRR method; (e) MSD method; (f) PCNN method; (g) PA-PCNN method; (h) proposed method

Fig. 8. Fusion results of different methods under “Steamboard” image. (a) IR image; (b) VIS image; (c) PCA method; (d) LatLRR method; (e) MSD method; (f) PCNN method; (g) PA-PCNN method; (h) proposed method