Fig. 1. Schematic diagram of CTC imaging system

Fig. 2. GAST-Net model. (a) Network model architecture; (b) GRU model structure; (c) attention mechanism diagram

Fig. 3. Flowchart of GAST-Net

Fig. 4. Comparison of structures predicted by two models and real reconstructed structure. (a) True three-dimensional structure; (b) 3D structure predicted by GAST-Net; (c) 3D structure predicted by CNN-LSTM; (d) 2D slices of real structure; (e) 2D slices of structure predicted by GAST-Net; (f) 2D slices of structure predicted by CNN-LSTM; (g) quality evaluation functions of two prediction structures and real structure

Fig. 5. Comparison of future moment prediction results from different methods. (a)‒(c) Real reconstruction results for future moments one, three, and five; (d)‒(f) reconstruction prediction results of GAST-Net for future moments one, three, and five; (g)‒(i) reconstruction prediction results of CNN-LSTM for future moments one, three, and five

Fig. 6. Cross sectional comparison and difference between combustion field and real data in the future. (a) The first moment of the future; (b) the third moment of the future; (c) the fifth moment of the future

Fig. 7. Quality evaluation functions between prediction results for multiple time steps and real data. (a) SSIM; (b) CORR; (c) RMSE; (d) PSNR

Fig. 8. Comparison of 3D structural views of predicted results and real data when inserting three frames. (a)‒(c) Inter-frame prediction reconstruction results; (d)‒(f) traditional reconstruction results; (g) quality evaluation functions of prediction and real structures

Fig. 9. Cross sectional comparison and difference of inter-frame predicted combustion field results and real data. (a) The first frame in the middle; (b) the second frame in the middle; (c) the third frame in the middle

Fig. 10. Quality evaluation functions for different number of inter-frame interpolations. (a) SSIM (b) CORR; (c) RMSE; (d) PSNR