Fig. 1. Overall structure block diagram.

Fig. 2. Image distortion caused by different turbulence intensities.

Fig. 3. Blur and distortion in images caused by varying turbulence intensities with an aerosol optical thickness of 0.01.

Fig. 4. Degradation effects of varying aerosol optical thicknesses on image quality. (1) Degraded images and PSF distributions for varying aerosol thicknesses at C_n²= 10⁻¹⁶ m^−2/3 turbulence intensity. (2) Degraded images and PSF distributions for varying aerosol thicknesses at C_n² = 10⁻¹⁵ m^−2/3 turbulence intensity.

Fig. 5. Field trials in the Jingyue region of Changchun City.

Fig. 6. Atmospheric refractive index structure constants at different times of the day. (a) July 30, 2024. (b) August 1, 2024.

Fig. 7. Aerosol extinction coefficient data. (a) Aerosol extinction coefficients at different heights on July 30, 2024, at 08:00, 11:00, 12:00, 13:00, 14:00, 15:00, and 18:00. (b) Aerosol extinction coefficients were measured at 0.06, 0.3, and 0.6 km on July 30, 2024, at various times. (c) Aerosol extinction coefficients at different heights on August 1, 2024, at 08:00, 11:00, 12:00, 13:00, 14:00, 15:00, and 18:00. (d) Aerosol extinction coefficients were measured at 0.06, 0.3, and 0.6 km on August 1, 2024, at various times.

Fig. 8. Comparison of measured and simulated degraded images of streetlights in the Jingyue region. (a) Simulation of atmospheric parameters at different times. (b) Comparison of measured and simulated degraded images with high turbulence intensity and aerosol thickness.

Fig. 9. Comparison of measured and simulated degraded images of rooftops in the Jingyue region. (a) Simulation of atmospheric parameters at different times. (b) Comparison of measured and simulated degraded images with high turbulence intensity and aerosol thickness.

Fig. 10. Quality evaluation of degraded images.

Fig. 11. Generator diagram.

Fig. 12. Multi-scale feature extraction.

Fig. 13. Discriminator structure diagram.

Fig. 14. Dataset preprocessing.

Fig. 15. Comparison of experimental results of different algorithms.

Fig. 16. Recovery effects of real atmospheric turbulence degradation images.