• Acta Optica Sinica
  • Vol. 39, Issue 3, 0301005 (2019)
Wei Tan*, Shixiang Cao, Wenwen Qi, and Hongyan He
Author Affiliations
  • Beijing Institute of Space Mechanics and Technology, Beijing 100094, China
  • show less
    DOI: 10.3788/AOS201939.0301005 Cite this Article Set citation alerts
    Wei Tan, Shixiang Cao, Wenwen Qi, Hongyan He. A Haze Removal Method for High-Resolution Remote Sensing Images[J]. Acta Optica Sinica, 2019, 39(3): 0301005 Copy Citation Text show less
    Schematic of triangulation interpolation
    Fig. 1. Schematic of triangulation interpolation
    Original images. (a) GF-2 panchromatic image; (b) GF-2 multispectral image; (c) Superview-1 panchromatic image; (d) Superview-1 multispectral image
    Fig. 2. Original images. (a) GF-2 panchromatic image; (b) GF-2 multispectral image; (c) Superview-1 panchromatic image; (d) Superview-1 multispectral image
    GF-2 images after dehazing process. (a) Haze image in blue spectral band; (b) image of bright object in red spectral band; (c) dark channel image; (d) non-dark channel image
    Fig. 3. GF-2 images after dehazing process. (a) Haze image in blue spectral band; (b) image of bright object in red spectral band; (c) dark channel image; (d) non-dark channel image
    Superview-1 images after dehazing process. (a) Haze image in blue spectral band; (b) image of bright object in red spectral band; (c) dark channel image; (d) non-dark channel image
    Fig. 4. Superview-1 images after dehazing process. (a) Haze image in blue spectral band; (b) image of bright object in red spectral band; (c) dark channel image; (d) non-dark channel image
    GF-2 images before and after dehazing processing. (a) Original panchromatic image; (b) panchromatic image after processing by Makarau's method; (c) panchromatic image after processing by proposed method; (d) original multispectral true color image; (e) true color image after processing by Makarau's method; (f) true color image after processing by proposed method
    Fig. 5. GF-2 images before and after dehazing processing. (a) Original panchromatic image; (b) panchromatic image after processing by Makarau's method; (c) panchromatic image after processing by proposed method; (d) original multispectral true color image; (e) true color image after processing by Makarau's method; (f) true color image after processing by proposed method
    Superview-1 images before and after dehazing processing. (a) Original panchromatic image; (b) panchromatic image after processing by Makarau's method; (c) panchromatic image after processing by proposed method; (d) original multispectral true color image; (e) true color image after processing by Makarau's method; (f) true color image after processing by proposed method
    Fig. 6. Superview-1 images before and after dehazing processing. (a) Original panchromatic image; (b) panchromatic image after processing by Makarau's method; (c) panchromatic image after processing by proposed method; (d) original multispectral true color image; (e) true color image after processing by Makarau's method; (f) true color image after processing by proposed method
    Comparison histogram of GF-2 image quality parameters before and after dehazing processing. (a) Average gray-scale gradient; (b) edge energy; (c) contrast; (d) sharpness
    Fig. 7. Comparison histogram of GF-2 image quality parameters before and after dehazing processing. (a) Average gray-scale gradient; (b) edge energy; (c) contrast; (d) sharpness
    Comparison histogram of Superview-1 image quality parameters before and after dehazing processing. (a) Average gray-scale gradient; (b) edge energy; (c) contrast; (d) sharpness
    Fig. 8. Comparison histogram of Superview-1 image quality parameters before and after dehazing processing. (a) Average gray-scale gradient; (b) edge energy; (c) contrast; (d) sharpness
    Schematics of color of local bright object before and after dehazing processing. (a) Original image; (b) image processed by Makarau's method; (c) image processed by proposed method
    Fig. 9. Schematics of color of local bright object before and after dehazing processing. (a) Original image; (b) image processed by Makarau's method; (c) image processed by proposed method
    Dehazing processing of whole scene image. (a) Original image and some details; (b) dehazing image and some details
    Fig. 10. Dehazing processing of whole scene image. (a) Original image and some details; (b) dehazing image and some details
    SatelliteLaunching timeGSD /mBand range /μm
    PAN: 0.8;PAN: 0.45-0.90;
    GF-22014-08-19MS: 3.2B1: 0.45-0.52;
    B2: 0.52-0.59;
    B3: 0.63-0.69;
    B4: 0.77-0.89
    PAN: 0.5;PAN: 0.45-0.90;
    Superview-12016-12-26MS: 2.0B1: 0.45-0.52;
    B2: 0.52-0.59;
    B3: 0.63-0.69;
    B4: 0.77-0.89
    Table 1. Satellite parameters
    Image quality parameterOriginal imageImage processed by Makarau's methodImage processed by proposed method
    Average gray-scale gradient0.46110.74801.0000
    Edge energy0.83970.95171.0000
    Contrast0.79330.94771.0000
    Sharpness0.44460.88821.0000
    Table 2. Comparison of quality parameters of GF-2 panchromatic images before and after dehazing processing
    Image quality parameterOriginal imageImage processed by Makarau's methodImage processed by proposed method
    Average gray-scale gradientB1: 0.3393;B2: 0.3754;B3: 0.2580B1: 0.7621;B2: 0.9283;B3: 0.7554B1: 1.0000;B2: 1.0000;B3: 1.0000
    Edge energyB1: 0.9398;B2: 0.9810;B3: 0.9702B1: 0.9709;B2: 0.9873;B3: 0.9893B1: 1.0000;B2: 1.0000;B3: 1.0000
    ContrastB1: 0.5625;B2: 0.5535;B3: 0.7573B1: 0.8493;B2: 0.8812;B3: 0.9058B1: 1.0000;B2: 1.0000;B3: 1.0000
    SharpnessB1: 0.6346;B2: 0.7083;B3: 0.4557B1: 0.8792;B2: 0.9102;B3: 0.7776B1: 1.0000;B2: 1.0000;B3: 1.0000
    Table 3. Comparison of quality parameters of GF-2 multispectral images before and after dehazing processing
    Image quality parameterOriginal imageImage processed by Makarau's methodImage processed by proposed method
    Average gray-scale gradient0.65260.81901.0000
    Edge energy0.96200.98711.0000
    Contrast0.88540.93831.0000
    Sharpness0.80400.92341.0000
    Table 4. Comparison of quality parameters of Superview-1 panchromatic images before and after dehazing processing
    Image quality parameterOriginal imageImage processed by Makarau's methodImage processed by proposed method
    Average gray-scale gradientB1: 0.5377;B2: 0.5766;B3: 0.6233B1: 0.8528;B2: 0.8929;B3: 0.8547B1: 1.0000;B2: 1.0000;B3: 1.0000
    Edge energyB1: 0.9435;B2: 0.9624;B3: 0.9773B1: 0.9772;B2: 0.9839;B3: 0.9959B1: 1.0000;B2: 1.0000;B3: 1.0000
    ContrastB1: 0.4369;B2: 0.6321;B3: 0.7657B1: 0.8645;B2: 0.9115;B3: 0.9198B1: 1.0000;B2: 1.0000;B3: 1.0000
    SharpnessB1: 0.7961;B2: 0.8046;B3: 0.8025B1: 0.9228;B2: 0.9388;B3: 0.9387B1: 1.0000;B2: 1.0000;B3: 1.0000
    Table 5. Comparison of quality parameters of Superview-1 multispectral spectral images before and after dehazing processing
    Wei Tan, Shixiang Cao, Wenwen Qi, Hongyan He. A Haze Removal Method for High-Resolution Remote Sensing Images[J]. Acta Optica Sinica, 2019, 39(3): 0301005
    Download Citation