# What Can Multifractal Analysis Tell Us about Hyperspectral Imagery?

^{1}

^{2}

^{*}

## Abstract

**:**

## 1. Introduction

#### 1.1. Fractals

#### 1.2. Multifractals

## 2. Data

## 3. Methodology

## 4. Results

#### 4.1. Degree of Multifractality for Radiance

#### 4.2. Degree of Multifractality for Reflectance

## 5. Discussion

#### 5.1. Interpretation of the Multifractal Results

#### 5.1.1. Influence of Atmospheric Absorption

_{2}absorption. In the NIR zone, three subzones can be clearly observed (bands 41–63, 64–84 and 85–103). This division may be caused by the absorbance of solar radiation by H

_{2}O vapour at the edges. The values for mountains reveal two fragments where there is a sudden increment (bands 63–65 and 81–86). These parts of the electromagnetic spectrum are where absorption caused by H

_{2}O is observed. Atmospheric correction causes smoothing of the Δ curve for the mountain landscape in these bands. It could suggest that the Δ parameter may be used to evaluate atmospheric correction quality. Moreover, in the bands adjacent to these zones (in SWIR 1 and SWIR 2), we observed increased values only for agriculture and mountain landscapes. It may be related to the fact that these two landscape samples include plants. After atmospheric correction, Δ for these bands also seems to be smoothed.

#### 5.1.2. Landscape Types and Dimensionality Reduction

#### 5.1.3. Size of Images and Spatial Resolution

#### 5.2. Comparison with Other Characteristics

#### 5.2.1. Correlation with Statistical Moments

^{2}) to evaluate the complementarity between Δ and statistical moments. In the first assessment, values of Δ for all bands for the four landscape types in reflectance images were combined. Low values of r

^{2}were obtained with four moments: mean (0.003), standard deviation (0.225), skewness (0.114) and kurtosis (0.034). In the next step, each landscape type was analysed separately (Table 6). In this case, we observed the highest coefficient values for urban landscape, notably for the mean (0.834) and skewness (0.858). This may be because the urban sample has the most complex structure and contains a dense mixture of dark and bright objects (see Figure 1c). Coefficient values were much lower for the three other landscape types.

#### 5.2.2. Comparison with Fractal Dimension

## 6. Conclusions

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Appendix A

Landscape Type | Original Image (Radiance) | Original Image (Reflectance) |
---|---|---|

Agriculture | f130522t01p00r11rdn_e | f130522t01p00r11_refl |

Mountains | f130522t01p00r12rdn_e | f130522t01p00r12_refl |

Urban | f130612t01p00r11rdn_e | f130612t01p00r11_refl |

Water | f130606t01p00r10rdn_e | f130606t01p00r10_refl |

**Figure A1.**Values of first four statistical moments (from top: mean, variance, skewness and kurtosis) determined for radiance (left column) and reflectance (right column) data for four landscape types: water (blue), mountains (black), agriculture (green) and urban (red).

## Appendix B

**Table A2.**Values of the Δ estimation for radiance and reflectance data. Bold font indicates the lowest and the highest Δ values for specific landscape types.

Radiance | Reflectance | ||||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|

VIS | NIR | A 1 | SWIR 1 | A 2 | SWIR 2 | All | VIS | NIR | SWIR 1 | SWIR 2 | All | ||

Agriculture | Min | 0.006 | 0.039 | 0.047 | 0.068 | 0.075 | 0.157 | 0.006 | 0.032 | 0.039 | 0.068 | 0.157 | 0.032 |

Mean | 0.088 | 0.070 | 0.073 | 0.106 | 0.197 | 0.264 | 0.137 | 0.151 | 0.072 | 0.101 | 0.262 | 0.151 | |

Max | 0.210 | 0.109 | 0.159 | 0.188 | 0.423 | 0.498 | 0.498 | 0.497 | 0.110 | 0.171 | 0.497 | 0.497 | |

Mountains | Min | 0.001 | 0.024 | 0.033 | 0.052 | 0.164 | 0.095 | 0.001 | 0.031 | 0.025 | 0.053 | 0.096 | 0.025 |

Mean | 0.031 | 0.037 | 0.116 | 0.070 | 0.349 | 0.135 | 0.092 | 0.100 | 0.034 | 0.064 | 0.129 | 0.079 | |

Max | 0.084 | 0.076 | 0.222 | 0.138 | 0.581 | 0.278 | 0.581 | 0.206 | 0.047 | 0.083 | 0.208 | 0.208 | |

Urban | Min | 0.002 | 0.034 | 0.024 | 0.071 | 0.032 | 0.082 | 0.002 | 0.045 | 0.041 | 0.058 | 0.061 | 0.041 |

Mean | 0.060 | 0.048 | 0.083 | 0.087 | 0.048 | 0.134 | 0.080 | 0.126 | 0.056 | 0.089 | 0.134 | 0.098 | |

Max | 0.104 | 0.055 | 0.484 | 0.103 | 0.062 | 0.172 | 0.484 | 0.198 | 0.064 | 0.108 | 0.176 | 0.198 | |

Water | Min | 0.0003 | 0.004 | 0.004 | 0.004 | 0.011 | 0.001 | 0.0003 | 0.003 | 0.005 | 0.005 | 0.002 | 0.002 |

Mean | 0.002 | 0.005 | 0.081 | 0.005 | 0.079 | 0.007 | 0.014 | 0.005 | 0.006 | 0.006 | 0.006 | 0.006 | |

Max | 0.004 | 0.005 | 0.768 | 0.006 | 0.234 | 0.030 | 0.768 | 0.025 | 0.006 | 0.006 | 0.022 | 0.025 |

**Table A3.**Errors in the Δ estimation for radiance data. Bold font indicates the highest mean value within a zone; green represents the lowest mean error for a specific landscape type; and red, the highest.

VIS | NIR | A 1 | SWIR 1 | A 2 | SWIR 2 | All | ||
---|---|---|---|---|---|---|---|---|

Agriculture | Min | 0.0004 4% | 0.0031 8% | 0.0036 7% | 0.0051 7% | 0.0060 7% | 0.0125 8% | 0.0004 4% |

Mean | 0.0070 7% | 0.0056 8% | 0.0061 8% | 0.0083 8% | 0.0240 13% | 0.0245 9% | 0.0124 8% | |

Max | 0.0192 9% | 0.0086 8% | 0.0132 13% | 0.0161 9% | 0.0464 29% | 0.0581 12% | 0.0581 29% | |

Mountains | Min | 0.0001 5% | 0.0008 2% | 0.0009 2% | 0.0034 4% | 0.0065 4% | 0.0047 3% | 0.0001 2% |

Mean | 0.0026 9% | 0.0027 7% | 0.0061 5% | 0.0044 6% | 0.0244 7% | 0.0075 6% | 0.0059 7% | |

Max | 0.0068 15% | 0.0087 11% | 0.0150 8% | 0.0071 8% | 0.0458 9% | 0.0156 7% | 0.0458 15% | |

Urban | Min | 0.0001 8% | 0.0080 18% | 0.0030 8% | 0.0062 9% | 0.0012 4% | 0.0031 4% | 0.0001 4% |

Mean | 0.010417% | 0.010822% | 0.029219% | 0.016319% | 0.0096 20% | 0.024718% | 0.016019% | |

Max | 0.0158 24% | 0.0131 24% | 0.2524 52% | 0.0185 23% | 0.0167 28% | 0.0347 22% | 0.2524 52% | |

Water | Min | 0.0000 17% | 0.0006 16% | 0.0006 14% | 0.0005 13% | 0.0021 19% | 0.0001 10% | 0.0000 10% |

Mean | 0.0003 17% | 0.0008 17% | 0.0224 24% | 0.0009 16% | 0.028833% | 0.0012 17% | 0.0039 18% | |

Max | 0.0006 23% | 0.0008 17% | 0.2124 34% | 0.0009 17% | 0.0846 46% | 0.0083 28% | 0.2124 46% | |

All | Mean | 0.0052 13% | 0.0050 14% | 0.0159 14% | 0.0075 12% | 0.0217 18% | 0.0145 12% |

**Table A4.**Modulus of the difference between error values of Δ calculated for reflectance and radiance data. Average values.

Landscape Type | VIS | NIR | SWIR 1 | SWIR 2 |
---|---|---|---|---|

Agriculture | 0.007 | 0.0003 | 0.001 | 0.001 |

304% | 5% | 5% | 2% | |

Mountains | 0.004 | 0.001 | 0.001 | 0.001 |

542% | 32% | 12% | 7% | |

Urban | 0.009 | 0.003 | 0.003 | 0.002 |

494% | 32% | 21% | 6% | |

Water | 0.001 | 0.0002 | 0.0001 | 0.00005 |

440% | 25% | 11% | 6% | |

All | 0.005 | 0.001 | 0.001 | 0.001 |

442% | 21% | 12% | 5% |

## Appendix C

**Figure A2.**Fractal dimension determined for radiance images of water (blue), mountain (black), agriculture (green) and urban (red) landscapes.

**Figure A3.**Fractal dimension determined for reflectance images of water (blue), mountain (black), agriculture (green) and urban (red) landscapes.

## References

- Mandelbrot, B.B. Fractals: Form, Chance, and Dimension; W.H. Freeman & Company: San Francisco, CA, USA, 1977; ISBN 0716704730. [Google Scholar]
- Mandelbrot, B.B. The Fractal Geometry of Nature; Einaudi Paperbacks; Henry Holt and Company: New York, NY, USA, 1983; ISBN 9780716711865. [Google Scholar]
- Chaudhuri, B.B.; Sarkar, N. Texture Segmentation Using Fractal Dimension. IEEE Trans. Pattern Anal. Mach. Intell.
**1995**, 17, 72–77. [Google Scholar] [CrossRef] - Gao, Q.; Zribi, M.; Escorihuela, J.M.; Baghdadi, N.; Segui, Q.P. Irrigation Mapping Using Sentinel-1 Time Series at Field Scale. Remote Sens.
**2018**, 10, 1495. [Google Scholar] [CrossRef] [Green Version] - Di Martino, G.; Di Simone, A.; Riccio, D. Fractal-Based Local Range Slope Estimation from Single SAR Image with Applications to SAR Despeckling and Topographic Mapping. Remote Sens.
**2018**, 10, 1294. [Google Scholar] [CrossRef] [Green Version] - Di Martino, G.; Iodice, A.; Riccio, D.; Ruello, G.; Zinno, I. The Role of Resolution in the Estimation of Fractal Dimension Maps From SAR Data. Remote Sens.
**2018**, 10, 9. [Google Scholar] [CrossRef] [Green Version] - Zhou, Y.; Lin, C.; Wang, S.; Liu, W.; Tian, Y. Estimation of Building Density with the Integrated Use of GF-1 PMS and Radarsat-2 Data. Remote Sens.
**2016**, 8, 969. [Google Scholar] [CrossRef] [Green Version] - Chen, Q.; Li, L.; Xu, Q.; Yang, S.; Shi, X.; Liu, X. Multi-Feature Segmentation for High-Resolution Polarimetric SAR Data Based on Fractal Net Evolution Approach. Remote Sens.
**2017**, 9, 570. [Google Scholar] [CrossRef] [Green Version] - Peleg, S.; Naor, J.; Hartley, R.; Avnir, D. Multiple Resolution Texture Analysis and Classification. IEEE Trans. Pattern Anal. Mach. Intell.
**1984**, PAMI-6, 518–523. [Google Scholar] [CrossRef] [Green Version] - Clarke, K.C. Computation of the Fractal Dimension of Topographic Surfaces Using the Triangular Prism Surface Area Method. Comput. Geosci.
**1986**, 12, 713–722. [Google Scholar] [CrossRef] - Lam, N.S.N.; De Cola, L. Fractals in Geography; Prentice Hall: Upper Saddle River, NJ, USA, 1993; ISBN 9780131058675. [Google Scholar]
- Sevcik, C. A Procedure to Estimate the Fractal Dimension of Waveforms. Complex. Int.
**1998**, 5, 1–19. [Google Scholar] - Turcotte, D.L. Fractals and Chaos in Geology and Geophysics, 2nd ed.; Cambridge University Press: Cambridge, UK, 1997. [Google Scholar]
- Mukherjee, K.; Ghosh, J.K.; Mittal, R.C. Variogram Fractal Dimension Based Features for Hyperspectral Data Dimensionality Reduction. J. Indian Soc. Remote Sens.
**2013**, 41, 249–258. [Google Scholar] [CrossRef] - Ghosh, J.K.; Somvanshi, A. Fractal-based Dimensionality Reduction of Hyperspectral Images. J. Indian Soc. Remote Sens.
**2008**, 36, 235–241. [Google Scholar] [CrossRef] - Aleksandrowicz, S.; Wawrzaszek, A.; Drzewiecki, W.; Krupinski, M. Change Detection Using Global and Local Multifractal Description. IEEE Geosci. Remote Sens. Lett.
**2016**, 13, 1183–1187. [Google Scholar] [CrossRef] - Dong, P. Fractal Signatures for Multiscale Processing of Hyperspectral Image Data. Adv. Space Res.
**2008**, 41, 1733–1743. [Google Scholar] [CrossRef] - Mukherjee, K.; Bhattacharya, A.; Ghosh, J.K.; Arora, M.K. Comparative Performance of Fractal Based and Conventional Methods for Dimensionality Reduction of Hyperspectral Data. Opt. Lasers Eng.
**2014**, 55, 267–274. [Google Scholar] [CrossRef] - Mukherjee, K.; Ghosh, J.K.; Mittal, R.C. Dimensionality Reduction of Hyperspectral Data Using Spectra Fractal Feature. Geocarto Int.
**2012**, 27, 515–531. [Google Scholar] [CrossRef] - Qiu, H.-I.; Lam, N.; Quattrochi, D.; Gamon, J. Fractal Characterization of Hyperspectral Imagery. Photogramm. Eng. Remote Sens.
**1999**, 65, 63–71. [Google Scholar] - Myint, S.W. Fractal Approaches in Texture Analysis and Classification of Remotely Sensed Data: Comparison with Spatial Autocorrelation Techniques and Simple Descriptive Statistics. Int. J. Remote Sens.
**2003**, 24, 1925–1947. [Google Scholar] [CrossRef] - Krupinski, M.; Wawrzaszek, A.; Drzewiecki, W.; Aleksandrowicz, S. Usefulness of the Fractal Dimension in the Context of Hyperspectral Data Description. In Proceedings of the 14th SGEM GeoConference on Informatics, Geoinformatics and Remote Sensing; STEF92 Technology, Albena, Bulgaria, 17–26 June 2014; Volume 3, pp. 367–374. [Google Scholar] [CrossRef]
- Sarkar, N.; Chaudhuri, B.B. An efficient differential box-counting approach to compute fractal dimension of image. IEEE Trans. Syst. Man Cybern.
**1994**, 24, 115–120. [Google Scholar] [CrossRef] [Green Version] - Sun, W.; Xu, G.; Gong, P.; Liang, S. Fractal analysis of remotely sensed images: A review of methods and applications. Int. J. Remote Sens.
**2006**, 27, 4963–4990. [Google Scholar] [CrossRef] - Wawrzaszek, A.; Aleksandrowicz, S.; Krupiski, M.; Drzewiecki, W. Influence of Image Filtering on Land Cover Classification when using Fractal and Multifractal Features. Photogramm. Fernerkund. Geoinf.
**2014**, 2014, 101–115. [Google Scholar] [CrossRef] - Drzewiecki, W.; Wawrzaszek, A.; Krupinski, M.; Aleksandrowicz, S.; Bernat, K. Comparison of selected textural features as global content-based descriptors of VHR satellite image—The EROS—A study 2013. In Proceedings of the 2013 Federated Conference on Computer Science and Information Systems, Krakow, Poland, 8–11 September 2013; pp. 43–49. [Google Scholar]
- Halsey, T.C.; Jensen, M.H.; Kadanoff, L.P.; Procaccia, I.; Shraiman, B.I. Fractal measures and their singularities: The characterization of strange sets. Nucl. Phys. B Proc. Suppl.
**1987**, 2, 501–511. [Google Scholar] [CrossRef] - Hentschel, H.G.E.; Procaccia, I. The infinite number of generalized dimensions of fractals and strange attractors. Phys. D Nonlinear Phenom.
**1983**, 8, 435–444. [Google Scholar] [CrossRef] - Su, H.; Sheng, Y.; Du, P. A New Band Selection Algorithm for Hyperspectral Data Based on Fractal Dimension. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci. Beijing
**2008**, XXXVII, 279–284. [Google Scholar] - Combrexelle, S.; Wendt, H.; Tourneret, J.-Y.; Mclaughlin, S.; Abry, P. Hyperspectral Image Analysis Using Multifractal Attributes. In Proceedings of the 7th IEEE Workshop on Hyperspectral Image and SIgnal Processing: Evolution in Remote Sensing (WHISPERS 2015), Tokyo, Japan, 2–5 June 2015; pp. 1–4. [Google Scholar] [CrossRef] [Green Version]
- Aleksandrowicz, S.; Wawrzaszek, A.; Jenerowicz, M.; Drzewiecki, W.; Krupinski, M. Local Multifractal Description of Bi-Temporal VHR Images. In Proceedings of the 2019 10th International Workshop on the Analysis of Multitemporal Remote Sensing Images (MultiTemp), Shanghai, China, 5–7 August 2019; pp. 1–3. [Google Scholar] [CrossRef]
- Grazzini, J.; Turiel, A.; Yahia, H.; Herlin, I.; Rocquencourt, I. Edge-preserving smoothing of high-resolution images with a partial multifractal reconstruction scheme. In Proceedings of the ISPRS 2004—International Society for Photogrammetry and Remote Sensing XXXV, Istambul, Turkey, 12–23 July 2004; pp. 1125–1129. [Google Scholar]
- Hu, M.-G.; Wang, J.-F.; Ge, Y. Super-resolution reconstruction of remote sensing images using multifractal analysis. Sensors
**2009**, 9, 8669–8683. [Google Scholar] [CrossRef] - Chen, Z.; Hu, Y.; Zhang, Y. Effects of Compression on Remote Sensing Image Classification Based on Fractal Analysis. IEEE Trans. Geosci. Remote Sens.
**2019**, 57, 4577–4590. [Google Scholar] [CrossRef] - Drzewiecki, W.; Wawrzaszek, A.; Aleksandrowicz, S.; Krupinski, M.; Bernat, K. Comparison of selected textural features as global content-based descriptors of VHR satellite image. In Proceedings of the 2013 IEEE International Geoscience and Remote Sensing Symposium—IGARSS, Melbourne, Australia, 21–26 July 2013; pp. 4364–4366. [Google Scholar] [CrossRef]
- Kupidura, P. The Comparison of Different Methods of Texture Analysis for Their Efficacy for Land Use Classification in Satellite Imagery. Remote Sens.
**2019**, 11, 1233. [Google Scholar] [CrossRef] [Green Version] - Chen, Q.; Zhao, Z.; Jiang, Q.; Zhou, J.-X.; Tian, Y.; Zeng, S.; Wang, J. Detecting subtle alteration information from ASTER data using a multifractal-based method: A case study from Wuliang Mountain, SW China. Ore Geol. Rev.
**2019**, 115, 103182. [Google Scholar] [CrossRef] - Ghosh, J.K.; Somvanshi, A.; Mittal, R.C. Fractal Feature for Classification of Hyperspectral Images of Moffit Field, USA. Curr. Sci.
**2008**, 94, 356–358. [Google Scholar] - Junying, S.; Ning, S. A Dimensionality Reduction Algorithm of Hyper Spectral Image Based on Fract Analysis. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci.
**2008**, XXXVII, 297–302. [Google Scholar] - Ziyong, Z. Multifractal Based Hyperion Hyperspectral Data Mining. In Proceedings of the 2010 Seventh International Conference on Fuzzy Systems and Knowledge Discovery, Yantai, China, 10–12 August 2010; Volume 5, pp. 2109–2113. [Google Scholar]
- Hosseini, A.; Ghassemian, H. Classification of Hyperspectral and Multifractal Images by Using Fractal Dimension of Spectral Response Curve. In Proceedings of the 20th Iranian Conference on Electrical Engineering (ICEE2012), Tehran, Iran, 15–17 May 2012; pp. 1452–1457. [Google Scholar] [CrossRef]
- Li, N.; Zhao, H.; Huang, P.; Jia, G.R.; Bai, X. A novel logistic multi-class supervised classification model based on multi-fractal spectrum parameters for hyperspectral data. Int. J. Comput. Math.
**2015**, 92, 836–849. [Google Scholar] [CrossRef] - Wan, X.; Zhao, C.; Wang, Y.; Liu, W. Stacked sparse autoencoder in hyperspectral data classification using spectral-spatial, higher order statistics and multifractal spectrum features. Infrared Phys. Technol.
**2017**, 86, 77–89. [Google Scholar] [CrossRef] - Krupiński, M.; Wawrzaszek, A.; Drzewiecki, W.; Aleksandrowicz, S.; Jenerowicz, M. Multifractal Parameters for Spectral Profile Description. In Proceedings of the IGARSS 2019—2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 28 July–2 August 2019; pp. 1256–1259. [Google Scholar] [CrossRef]
- Jenerowicz, M.; Wawrzaszek, A.; Krupiński, M.; Drzewiecki, W.; Aleksandrowicz, S. Aplicability of Multifractal Features as Descriptors of the Complex Terrain Situation in IDP/Refugee Camps. In Proceedings of the IGARSS 2019—2019 IEEE International Geoscience and Remote Sensing Symposium, Yokohama, Japan, 28 July–2 August 2019; pp. 2662–2665. [Google Scholar] [CrossRef]
- Jenerowicz, M.; Wawrzaszek, A.; Drzewiecki, W.; Krupiński, M.; Aleksandrowicz, S. Multifractality in Humanitarian Applications: A Case Study of Internally Displaced Persons/Refugee Camps. IEEE J. Sel. Top. Appl. Earth Obs. Remote Sens.
**2019**, 1–8. [Google Scholar] [CrossRef] - Drzewiecki, W.; Wawrzaszek, A.; Krupiński, M.; Aleksandrowicz, S.; Bernat, K. Applicability of multifractal features as global characteristics of WorldView-2 panchromatic satellite images. Eur. J. Remote Sens.
**2016**, 49, 809–834. [Google Scholar] [CrossRef] [Green Version] - Wawrzaszek, A.; Walichnowska, M.; Krupiński, M. Evaluation of degree of multifractality for description of high resolution data aquired by Landsat satellites. Arch. Fotogram. Kartogr. Teledetekcji
**2015**, 27, 175–184. (In Polish) [Google Scholar] [CrossRef] - Green, R.O.; Eastwood, M.L.; Sarture, C.M.; Chrien, T.G.; Aronsson, M.; Chippendale, B.J.; Faust, J.A.; Pavri, B.E.; Chovit, C.J.; Solis, M.; et al. Imaging spectroscopy and the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS). Remote Sens. Environ.
**1998**, 65, 227–248. [Google Scholar] [CrossRef] - Gao, B.-C.; Heidebrecht, K.B.; Goetz, A.F.H. Derivation of scaled surface reflectances from AVIRIS data. Remote Sens. Environ.
**1993**, 44, 165–178. [Google Scholar] [CrossRef] - Gao, B.-C.; Davis, C.O. Development of a line-by-line-based atmosphere removal algorithm for airborne and spaceborne imaging spectrometers. In Proceedings of the Imaging Spectrometry III, San Diego, CA, USA, 31 October 1997; Volume 3118. [Google Scholar]
- Wawrzaszek, A.; Krupinski, M.; Aleksandrowicz, S.; Drzewiecki, W. Fractal and multifractal characteristics of very high resolution satellite images. In Proceedings of the 2013 IEEE International Geoscience and Remote Sensing Symposium—IGARSS, Melbourne, Australia, 21–26 July 2013; pp. 1501–1504. [Google Scholar] [CrossRef]
- Wawrzaszek, A.; Echim, M.; Bruno, R. Multifractal analysis of heliospheric magnetic field fluctuations observed by Ulysses. Astrophys. J.
**2019**, 876, 153. [Google Scholar] [CrossRef] [Green Version]

**Figure 1.**Selected spectral bands of four Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) images used in the analyses: (

**a**) agriculture, (

**b**) mountains, (

**c**) urban and (

**d**) water.

**Figure 2.**Degree of multifractality (Δ) determined for radiance images of water (blue), mountain (black), agriculture (green) and urban (red) landscapes. The whole spectrum is divided according to ranges presented in Table 3.

**Figure 3.**Degree of multifractality calculated for radiance and reflectance images of water (blue), mountain (black), agriculture (green) and urban (red) landscapes, divided into four zones according to Table 3.

**Figure 4.**Degree of multifractality calculated for radiance images of different sizes: 512 × 512, 256 × 256, 128 × 128 and 64 × 64 pixels, and four landscape types: agriculture (green), mountains (black), urban (red) and water (blue).

**Table 1.**Examples of local and global (multi)fractal descriptions applied to hyperspectral data analysis.

Paper | Sensor/Dataset | Number of Bands | Image Size | Parameters Used/Method | |
---|---|---|---|---|---|

Global description | Qiu et al. (1999) [20] | AVIRIS ^{1}/MalibuAVIRIS/LA | 224 224 | 614 × 512 614 × 512 | Fractal Dimension/isarithm method and triangular prism method |

Myint et al. (2003) [21] | ATLAS (5 classes) | 15 | 17 × 17 33 × 32 65 × 65 | Fractal dimension/isarithm, triangular prism and variogram method | |

Su et al. (2008) [29] | OMIS ^{2}/Beijing | 64 | 536 × 512 | Fractal dimension/double blanket method | |

Krupiński et al. (2014) [22] | AVIRIS (4 classes) AVIRIS/Malibu AVIRIS/LA | 224 224 224 | 512 × 512 512 × 512 512 × 512 | Fractal dimension/differential box counting method | |

Local description | Combrexelle et al. (2015) [30] | Hyspex/Madonna AVIRIS/Moffit Field | 160 224 | 256 × 256 64 × 64 16 × 16 | Coefficients of the polynomial describing multifractal spectrum/wavelet leader multifractal formalism |

^{1}Airborne Visible/Infrared Imaging Spectrometer,

^{2}Operational Modular Imaging Spectrometer.

**Table 2.**Examples of (multi)fractal spectral curve descriptions applied to hyperspectral data analysis.

Paper | Sensor/Dataset | No. of Bands | No. of Classes | Parameters Used/Method | |
---|---|---|---|---|---|

Spectral Curve Description | Dong et al. (2008) [17] | HYPERION | 138 of 242 | 5 | Fractal dimension/blanket method |

Ghosh et al. (2008) [15] | AVIRIS/Moffit Field | 224 | 4 | Fractal dimension/adapted Hausdorff metric | |

Ghosh et al. (2008) [38] | AVIRIS/Moffit Field | 30 128 | 5 | Fractal dimension/adapted Hausdorff metric | |

Junying et al. (2008) [39] | MAIS ^{1}OMIS | 176 of 220 | 4 4 | Fractal dimension/step measurement method | |

Ziyong et al. (2010) [40] | HYPERION | 191 of 210 12 | - | Fractal dimension/modified blanked method | |

Hosseini et al. (2012) [41] | HYDICE ^{2}/WashingtonF210 | 191 of 210 12 | 6 9 | Fractal dimension/Hausdorff metric | |

Mukherjee et al. (2012) [14] | HYDICE AVIRIS/Indian Pine AVIRIS/Cuprite | 188 of 210 200 of 224 197 of 224 | 5 9 of 16 14 | Fractal dimension/power spectrum method | |

Mukherjee et al. (2013) [19] | HYDICE AVIRIS/Indian Pine AVIRIS/Cuprite | 188 of 210 200 of 224 197 of 224 | 5 9 of 16 14 | Fractal dimension/variogram method | |

Mukherjee et al. (2014) [18] | AVIRIS/Indian Pine AVIRIS/Cuprite | 200 of 224 197 of 224 | 9 of 16 14 | Fractal dimension/Sevcik’s method, power spectrum method, variogram method | |

Li et al. (2015) [42] | PHI ^{3}/FangluAVIRIS/Indian Pines | 64 200 of 224 | 6 16 | 4 parameters related to multifractal spectrum | |

Wan et al. (2017) [43] | AVIRIS/Indian Pines AVIRIS/KSC | 200 of 224 176 of 224 | 9 of 16 13 | Holder exponent, multifractal spectrum features | |

Krupiński et al. (2019) [44] | CASI ^{4}/University of Houston | 144 | 15 | 6 parameters related to multifractal spectrum/multifractal detrended fluctuation analysis |

^{1}Modular Airborne Imaging Spectrometer,

^{2}Hyperspectral Digital Imagery Collection Experiment,

^{3}Pushbroom Hyperspectral Imager,

^{4}Compact Airborne Spectrographic Imager.

Zone Name | VIS | NIR | A 1 | SWIR 1 | A 2 | SWIR 2 |
---|---|---|---|---|---|---|

Bands | 1–40 | 41–103 | 104–114 | 115–152 | 153–168 | 169–224 |

Wavelength (nm) | 366–724 | 734–1313 | 1323–1423 | 1433–1802 | 1811–1937 | 1947–2496 |

**Table 4.**Modulus of the difference between Δ values for reflectance and radiance data. Average absolute and relative (%) values. Relative values refer to radiance data.

Landscape Type | VIS | NIR | SWIR 1 | SWIR 2 |
---|---|---|---|---|

Agriculture | 0.074 | 0.003 | 0.007 | 0.004 |

233% | 4% | 4% | 1% | |

Mountains | 0.062 | 0.005 | 0.004 | 0.009 |

524% | 11% | 8% | 5% | |

Urban | 0.063 | 0.007 | 0.000 | 0.002 |

383% | 15% | 5% | 2% | |

Water | 0.003 | 0.001 | 0.001 | 0.0003 |

446% | 25% | 11% | 6% | |

All | 0.049 | 0.005 | 0.004 | 0.004 |

394% | 9% | 7% | 3% |

**Table 5.**Modulus of the difference between values of Δ calculated for radiance data for original and resampled data. Average values.

Mean Δ Differences | Mean Error Differences | |||||
---|---|---|---|---|---|---|

Landscape Type | 256 × 256 | 128 × 128 | 64 × 64 | 256 × 256 | 128 × 128 | 64 × 64 |

Agriculture | 0.007 | 0.007 | 0.012 | 0.003 | 0.008 | 0.014 |

4% | 4% | 10% | 2% | 5% | 11% | |

Mountains | 0.006 | 0.012 | 0.015 | 0.001 | 0.002 | 0.002 |

6% | 14% | 18% | 1% | 2% | 2% | |

Urban | 0.023 | 0.033 | 0.037 | 0.008 | 0.012 | 0.014 |

29% | 44% | 52% | 4% | 7% | 8% | |

Water | 0.005 | 0.008 | 0.011 | 0.002 | 0.004 | 0.004 |

21% | 44% | 76% | 2% | 2% | 5% | |

All | 0.010 | 0.015 | 0.019 | 0.003 | 0.005 | 0.009 |

15% | 26% | 39% | 2% | 3% | 6% |

**Table 6.**Correlation coefficients (r

^{2}) between the degree of multifractality and four moments calculated for reflectance for the four landscape types after removal of outliers.

Landscape Type | Mean | Standard Deviation | Skewness | Kurtosis |
---|---|---|---|---|

Agriculture | 0.452 | 0.003 | 0.012 | 0.508 |

Mountains | 0.664 | 0.075 | 0.038 | 0.012 |

Urban | 0.834 | 0.500 | 0.858 | 0.762 |

Water | 0.255 | 0.180 | 0.039 | 0.008 |

Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |

© 2020 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Krupiński, M.; Wawrzaszek, A.; Drzewiecki, W.; Jenerowicz, M.; Aleksandrowicz, S.
What Can Multifractal Analysis Tell Us about Hyperspectral Imagery? *Remote Sens.* **2020**, *12*, 4077.
https://doi.org/10.3390/rs12244077

**AMA Style**

Krupiński M, Wawrzaszek A, Drzewiecki W, Jenerowicz M, Aleksandrowicz S.
What Can Multifractal Analysis Tell Us about Hyperspectral Imagery? *Remote Sensing*. 2020; 12(24):4077.
https://doi.org/10.3390/rs12244077

**Chicago/Turabian Style**

Krupiński, Michał, Anna Wawrzaszek, Wojciech Drzewiecki, Małgorzata Jenerowicz, and Sebastian Aleksandrowicz.
2020. "What Can Multifractal Analysis Tell Us about Hyperspectral Imagery?" *Remote Sensing* 12, no. 24: 4077.
https://doi.org/10.3390/rs12244077