Fig. 1. Schematic diagram of orthogonal color transfer model in orthogonal color space

Fig. 2. Numerically indistinguishable clusters are separated in deep feature space

Fig. 3. Flow chart of the proposed orthogonal color transfer network with deep feature fusion

Fig. 4. Deep features extracted by the pre-trained VGG16 network as front-end network

Fig. 5. Framework of fully connected network for deep feature fusion proposed for color transfer

Fig. 6. Experimental equipment. (a) Spectral transmission of optical filters; (b) filter wheel; (c) normalized spectral sensitivity of EMCCD camera

Fig. 7. Illustration of proposed network

Fig. 8. Test results on 16 source-reference image pairs

Fig. 9. Evaluation of predictions on 16 test image pairs

Fig. 10. Comparison of model prediction results

Fig. 11. Training loss curves of orthogonal color space ablation test

Fig. 12. Comparison of prediction results of ablation tests

Fig. 13. Part of comparative experimental results

Fig. 14. Color transfer results of EMCCD source images captured under illumination of 0.1--1 lx

Fig. 15. Color transfer result of the EMCCD source images captured under illumination of 0.01 lx