Fig. 1. Basic structure of the CUP-VISAR system for measuring shock wave velocity.

Fig. 2. Framework for coding matrix design utilizing a CVAE network.

Fig. 3. Network architecture diagram for encoding and decoding.

Fig. 4. Mask. (a) Random mask; (b) 300 epoch feature-free fusion mask; (c) 300 epoch feature fusion mask.

Fig. 5. Velocity fringe reconstruction of shock wave diagnosis by four algorithms with different masks. (a1) Random mask; (a2) measurement with the random mask; (a3)–(a6) 25th frame reconstruction of the random mask; (b1) feature-free fusion mask; (b2) measurement with the feature-free fusion mask; (a3)–(a6) 25th frame reconstruction of the feature-free fusion mask; (c1) feature fusion mask; (c2) measurement with feature fusion mask; (c3)–(c6) 25th frame reconstruction of the feature fusion mask.

Fig. 6. Experimental mask plate specification drawing. The coding-aperture ratio of the mask is 7:3. (a) Feature fusion mask; (b) random mask.

Fig. 7. Velocity fringe reconstruction performance in various coding aperture ratios. (a1), (b1) PSNR performance comparison for different algorithms; (a2), (b2) SSIM performance comparison for different algorithms.