Fig. 1. Flow chart of adaptive bi-channel prior image dehazing method based on superpixel

Fig. 2. Haze-free images， bright channel images and histogram statistics of bright channel values

Fig. 3. Extract white or bright colors and black or dark pixels from hazy and haze-free images

Fig. 4. Extraction results of white or bright and black or dark pixels in fog-free images and composite foggy images

Fig. 5. Histogram statistics and cumulative distribution diagram of error E_w value， E_b value and total error E value

Fig. 6. Influence of soft matting filter on the dehazing result obtained by estimating the transmittance map based on a fixed-size rectangular area

Fig. 7. Haze-free image and images based on dark channel in different regions

Fig. 8. DCP， pixel-based DCP and superpixel-based BiCP method to obtain dehazing results

Fig. 9. Transmittance map and atmospheric light map based on super pixel BiCP dehazing method

Fig. 10. Atmospheric light， transmittance and dehazing results based on the superpixel BiCP and ABiCP methods

Fig. 11. DCP， pixel-based DCP， BiCP and ABiCP methods images dehazing results

Fig. 12. Example images of objective evaluation index of "BIT01" dehazing results obtained by various dehazing methods

Fig. 13. Quantitative comparison of dehazing results on the FRIDA dataset