Fig. 1. Flow chart of time-domain iterative wavelet transform algorithm

Fig. 2. Diagrams of SSIM and time consumption changing with levels and iterations under different SNR datasets. (a)--(b) Under low and middle SNR datasets; (c)--(d) under high SNR datasets

Fig. 3. Different simulated SNR datasets and separated signals after processing with different algorithms. (a) Simulated SNR10 dataset; (b) separated signals with different algorithms for SNR10 dataset; (c) simulated SNR6 dataset; (d) separated signals with different algorithms for SNR6 dataset; (e) simulated SNR2 dataset; (f) separated signals with different algorithms for SNR2 dataset; (g) normalized intensity of single fluorescence signal

Fig. 4. Comparison of SSIM and PSNR under different algorithms in simulated datasets with high, middle and low SNRs and time consumption of different algorithms. (a) Comparison of SSIM and PSNR under different algorithms in SNR10 datasets; (b) comparison of SSIM and PSNR under different algorithms in SNR6 datasets; (c) comparison of SSIM and PSNR under different algorithms in SNR2 datasets; (d) time consumption of different algorithms

Fig. 5. Reconstructed super-resolved images of Tubulin ConjAL647. (a)--(f) Images reconstructed with TDIWT, EVER, GF, BEWT, RB, and MF algorithms, respectively; (g) raw dataset

Fig. 6. Three-dimensional experimental device diagram of easySTORM. (a) Three-dimensional diagram; (b) schematic

Fig. 7. Super-resolved images reconstructed with different algorithms. (a)--(f) Super-resolved images reconstructed with TDIWT, EVER, GF, BEWT, RB, and MF pre-processing; (g) bright-field fluorescence image; (h) structural noise in dashed box