# Unsupervised Parameterization for Optimal Segmentation of Agricultural Parcels from Satellite Images in Different Agricultural Landscapes

^{1}

^{2}

^{3}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Study Area and Data

## 3. Methodology

#### 3.1. Selection of the 21 Tiles

#### 3.2. Image Segmentation

#### 3.3. Segmentation Optimization

#### 3.3.1. Existing Unsupervised Segmentation Evaluation Metrics

#### 3.3.2. Metric Proposal Based on Absolute Difference (AD)

#### 3.3.3. Unsupervised Segmentation Optimization

- The domain space (minimum and maximum values) of each input parameter. The domain space of scale was defined as 20 and 200, for shape 0.0 and 0.9, and for compactness 0.0 and 1.0. These parameter ranges were also used by Tetteh et al. [47] in their approach.
- An objective function to optimize. For our study, the objective function to optimize is f(x), where x is a parameter combination of scale, shape, and compactness. The function takes the parameter combination, performs image segmentation, computes the GS of the segmentation output, and finally returns the GS.
- A surrogate model for the objective function. To build the surrogate model, one has to first define a prior probability distribution that captures the prior behavior of the objective function. We chose Gaussian Processes (GP) [50] as the prior probability distribution. Then, some initial parameter combinations together with their corresponding GS are used to initialize the whole optimization process. We used 125 parameter combinations as initialization samples. These 125 parameter combinations were selected in a way to ensure uniform and representative distribution over each parameter space. For scale, the values were (40, 80, 120, 160, 200), and for both shape and compactness, the values were (0.1, 0.3, 0.5, 0.7, 0.9). The grid search method was used to calculate the corresponding GS for the 125 samples. These samples were used to update the GP to obtain posterior probability distribution over the objective function.
- An acquisition function to be used in sampling new parameter combinations to be evaluated with the objective function. For the acquisition function, we used expected improvement (EI) [51]. EI is used to iteratively select new parameter combinations with the highest probability of optimizing the objection function. We sampled 50 new parameter combinations with the EI function in 50 iterations. At each iteration, out of 10,000 parameter combinations randomly sampled from the domain space, the combination with the highest likelihood of improving upon the current optimal parameter combination is identified by the EI function using the current posterior probability distribution. Then, this identified parameter combination is evaluated with the objective function, and the corresponding GS is used to update the current posterior probability distribution. In all, 175 combinations were used within the Bayesian optimization approach to identify the optimal one.

#### 3.4. Empirical Discrepancy Measures

## 4. Results

#### 4.1. Optimal Segmentation Based on Default Optimization

#### 4.2. Optimal Segmentation Based on Bayesian Optimization

## 5. Discussion

## 6. Conclusions

## Author Contributions

## Funding

## Acknowledgments

## Conflicts of Interest

## Appendix A

Tile | Image Date | Agric. Land Cover | No. of Land-Use Types | No. of LPIS Parcels | Min. Area (Ha) | Max. Area (Ha) | Mean Area (Ha) |
---|---|---|---|---|---|---|---|

T1 | 20 May 2018 | 62.29% | 12 | 1308 | 0.232 | 25.777 | 4.097 |

T2 | 5 May 2018 | 62.76% | 8 | 1344 | 0.173 | 21.726 | 2.398 |

T3 | 8 May 2018 | 80.91% | 4 | 2341 | 0.191 | 53.279 | 2.924 |

T4 | 7 May 2018 | 53.30% | 14 | 1671 | 0.169 | 35.281 | 2.522 |

T5 | 5 May 2018 | 76.83% | 14 | 1957 | 0.180 | 18.888 | 3.219 |

T6 | 5 May 2018 | 79.61% | 11 | 2500 | 0.168 | 22.639 | 2.565 |

T7 | 5 May 2018 | 68.08% | 16 | 2140 | 0.203 | 25.181 | 2.579 |

T8 | 8 May 2018 | 50.17% | 12 | 1100 | 0.199 | 30.562 | 3.704 |

T9 | 5 May 2018 | 70.43% | 11 | 1613 | 0.190 | 44.890 | 3.699 |

T10 | 5 May 2018 | 70.13% | 12 | 2441 | 0.177 | 26.253 | 2.243 |

T11 | 5 May 2018 | 71.41% | 15 | 1625 | 0.172 | 30.012 | 3.709 |

T12 | 5 May 2018 | 90.15% | 12 | 1798 | 0.171 | 50.494 | 3.127 |

T13 | 5 May 2018 | 92.31% | 12 | 1221 | 0.181 | 64.772 | 6.203 |

T14 | 5 May 2018 | 63.62% | 15 | 1894 | 0.176 | 26.646 | 2.637 |

T15 | 5 May 2018 | 36.63% | 15 | 809 | 0.203 | 23.687 | 3.580 |

T16 | 5 May 2018 | 58.45% | 14 | 1752 | 0.181 | 29.022 | 2.781 |

T17 | 5 May 2018 | 61.10% | 14 | 1538 | 0.180 | 28.160 | 2.994 |

T18 | 5 May 2018 | 37.26% | 13 | 729 | 0.193 | 28.514 | 4.158 |

T19 | 7 May 2018 | 14.29% | 8 | 420 | 0.217 | 25.855 | 2.471 |

T20 | 7 May 2018 | 33.35% | 13 | 744 | 0.191 | 36.408 | 3.111 |

T21 | 7 May 2018 | 90.84% | 11 | 1340 | 0.213 | 62.730 | 5.883 |

**Table A2.**Empirical discrepancy measures computed for each optimal segmentation result identified by the AD and Böck metrics based on the default optimization (shape = 0.1, compactness = 0.5). The bold-faced texts within the body of the table are the optimal results.

Tile | Scale | Shape | Compactness | QR | OR | UR | RMS | Metric |
---|---|---|---|---|---|---|---|---|

T1 | 190 | 0.100 | 0.500 | 55.53% | 0.115 | 0.375 | 0.278 | AD |

T1 | 300 | 0.100 | 0.500 | 38.42% | 0.057 | 0.597 | 0.424 | Böck |

T2 | 80 | 0.100 | 0.500 | 36.94% | 0.334 | 0.467 | 0.406 | AD |

T2 | 70 | 0.100 | 0.500 | 36.07% | 0.387 | 0.427 | 0.407 | Böck |

T3 | 150 | 0.100 | 0.500 | 57.91% | 0.183 | 0.304 | 0.251 | Böck |

T3 | 140 | 0.100 | 0.500 | 57.80% | 0.192 | 0.296 | 0.250 | AD |

T4 | 200 | 0.100 | 0.500 | 28.33% | 0.121 | 0.685 | 0.492 | AD |

T4 | 280 | 0.100 | 0.500 | 20.84% | 0.076 | 0.779 | 0.553 | Böck |

T5 | 160 | 0.100 | 0.500 | 44.69% | 0.163 | 0.477 | 0.356 | AD |

T5 | 200 | 0.100 | 0.500 | 39.00% | 0.122 | 0.563 | 0.407 | Böck |

T6 | 170 | 0.100 | 0.500 | 42.45% | 0.169 | 0.502 | 0.375 | AD |

T6 | 180 | 0.100 | 0.500 | 41.24% | 0.161 | 0.520 | 0.385 | Böck |

T7 | 190 | 0.100 | 0.500 | 32.84% | 0.128 | 0.631 | 0.455 | AD |

T7 | 270 | 0.100 | 0.500 | 25.46% | 0.084 | 0.729 | 0.519 | Böck |

T8 | 120 | 0.100 | 0.500 | 48.77% | 0.274 | 0.339 | 0.308 | AD |

T8 | 170 | 0.100 | 0.500 | 41.78% | 0.150 | 0.513 | 0.378 | Böck |

T9 | 170 | 0.100 | 0.500 | 44.88% | 0.215 | 0.446 | 0.350 | AD |

T9 | 300 | 0.100 | 0.500 | 33.86% | 0.101 | 0.635 | 0.454 | Böck |

T10 | 180 | 0.100 | 0.500 | 36.66% | 0.143 | 0.584 | 0.425 | AD |

T10 | 210 | 0.100 | 0.500 | 33.06% | 0.126 | 0.631 | 0.455 | Böck |

T11 | 230 | 0.100 | 0.500 | 35.67% | 0.127 | 0.595 | 0.430 | AD |

T11 | 230 | 0.100 | 0.500 | 35.67% | 0.127 | 0.595 | 0.430 | Böck |

T12 | 150 | 0.100 | 0.500 | 40.77% | 0.209 | 0.504 | 0.386 | AD |

T12 | 270 | 0.100 | 0.500 | 27.78% | 0.126 | 0.697 | 0.501 | Böck |

T13 | 240 | 0.100 | 0.500 | 42.42% | 0.177 | 0.495 | 0.372 | AD |

T13 | 300 | 0.100 | 0.500 | 35.18% | 0.142 | 0.601 | 0.436 | Böck |

T14 | 160 | 0.100 | 0.500 | 36.03% | 0.162 | 0.574 | 0.422 | AD |

T14 | 280 | 0.100 | 0.500 | 21.74% | 0.077 | 0.768 | 0.546 | Böck |

T15 | 220 | 0.100 | 0.500 | 32.40% | 0.090 | 0.648 | 0.463 | AD |

T15 | 300 | 0.100 | 0.500 | 22.41% | 0.062 | 0.764 | 0.542 | Böck |

T16 | 180 | 0.100 | 0.500 | 38.46% | 0.114 | 0.566 | 0.408 | AD |

T16 | 280 | 0.100 | 0.500 | 26.20% | 0.064 | 0.723 | 0.513 | Böck |

T17 | 200 | 0.100 | 0.500 | 31.89% | 0.137 | 0.634 | 0.459 | AD |

T17 | 240 | 0.100 | 0.500 | 27.14% | 0.111 | 0.700 | 0.501 | Böck |

T18 | 200 | 0.100 | 0.500 | 47.11% | 0.167 | 0.434 | 0.329 | Böck |

T18 | 190 | 0.100 | 0.500 | 47.06% | 0.168 | 0.427 | 0.325 | AD |

T19 | 210 | 0.100 | 0.500 | 37.29% | 0.092 | 0.595 | 0.426 | AD |

T19 | 50 | 0.100 | 0.500 | 36.58% | 0.552 | 0.207 | 0.417 | Böck |

T20 | 220 | 0.100 | 0.500 | 29.21% | 0.123 | 0.676 | 0.486 | AD |

T20 | 260 | 0.100 | 0.500 | 27.79% | 0.102 | 0.698 | 0.499 | Böck |

T21 | 270 | 0.100 | 0.500 | 43.91% | 0.133 | 0.499 | 0.365 | AD |

T21 | 300 | 0.100 | 0.500 | 40.52% | 0.117 | 0.546 | 0.395 | Böck |

**Table A3.**Empirical discrepancy measures computed for the unsupervised Bayesian optimization approaches based on the Böck and AD metrics, and the supervised Bayesian optimization approach (SUP) that was used to maximize the QR measure. The bold-faced texts within the body of the table are the optimal results.

Tile | Scale | Shape | Compactness | QR | OR | UR | RMS | Metric |
---|---|---|---|---|---|---|---|---|

T1 | 51 | 0.900 | 0.966 | 69.17% | 0.117 | 0.224 | 0.178 | SUP |

T1 | 160 | 0.300 | 0.500 | 57.47% | 0.126 | 0.349 | 0.263 | AD |

T1 | 200 | 0.841 | 0.917 | 34.39% | 0.035 | 0.648 | 0.459 | Böck |

T2 | 40 | 0.900 | 0.300 | 42.04% | 0.219 | 0.479 | 0.372 | SUP |

T2 | 42 | 0.792 | 0.176 | 40.28% | 0.309 | 0.429 | 0.374 | AD |

T2 | 56 | 0.415 | 0.192 | 37.40% | 0.402 | 0.395 | 0.398 | Böck |

T3 | 77 | 0.842 | 0.906 | 68.46% | 0.117 | 0.235 | 0.186 | SUP |

T3 | 117 | 0.420 | 1.000 | 62.79% | 0.164 | 0.263 | 0.219 | AD |

T3 | 138 | 0.279 | 0.175 | 59.14% | 0.165 | 0.304 | 0.245 | Böck |

T4 | 34 | 0.900 | 0.410 | 50.84% | 0.290 | 0.297 | 0.293 | SUP |

T4 | 116 | 0.655 | 1.000 | 38.04% | 0.121 | 0.576 | 0.416 | AD |

T4 | 174 | 0.666 | 0.753 | 24.88% | 0.076 | 0.738 | 0.524 | Böck |

T5 | 42 | 0.900 | 0.783 | 58.78% | 0.205 | 0.273 | 0.242 | SUP |

T5 | 132 | 0.468 | 0.701 | 47.21% | 0.149 | 0.459 | 0.341 | AD |

T5 | 162 | 0.395 | 0.452 | 42.52% | 0.124 | 0.524 | 0.381 | Böck |

T6 | 40 | 0.900 | 0.500 | 57.67% | 0.225 | 0.269 | 0.248 | SUP |

T6 | 127 | 0.422 | 0.083 | 46.98% | 0.172 | 0.442 | 0.335 | AD |

T6 | 144 | 0.377 | 0.000 | 46.05% | 0.161 | 0.466 | 0.348 | Böck |

T7 | 40 | 0.900 | 0.500 | 55.70% | 0.209 | 0.307 | 0.263 | SUP |

T7 | 183 | 0.088 | 0.401 | 35.14% | 0.142 | 0.601 | 0.436 | AD |

T7 | 178 | 0.686 | 0.611 | 29.39% | 0.071 | 0.692 | 0.492 | Böck |

T8 | 46 | 0.853 | 0.665 | 56.91% | 0.261 | 0.240 | 0.251 | SUP |

T8 | 120 | 0.100 | 0.300 | 49.20% | 0.261 | 0.339 | 0.303 | AD |

T8 | 160 | 0.300 | 0.100 | 43.71% | 0.145 | 0.499 | 0.367 | Böck |

T9 | 56 | 0.900 | 0.548 | 56.93% | 0.191 | 0.310 | 0.258 | SUP |

T9 | 129 | 0.398 | 1.000 | 49.61% | 0.212 | 0.384 | 0.310 | AD |

T9 | 200 | 0.300 | 0.500 | 41.28% | 0.148 | 0.532 | 0.390 | Böck |

T10 | 40 | 0.900 | 0.700 | 54.15% | 0.196 | 0.336 | 0.275 | SUP |

T10 | 189 | 0.000 | 0.380 | 37.43% | 0.152 | 0.573 | 0.419 | AD |

T10 | 184 | 0.587 | 0.633 | 33.58% | 0.084 | 0.641 | 0.457 | Böck |

T11 | 50 | 0.900 | 0.699 | 58.31% | 0.200 | 0.277 | 0.241 | SUP |

T11 | 200 | 0.100 | 0.900 | 40.52% | 0.143 | 0.528 | 0.386 | AD |

T11 | 108 | 0.900 | 0.777 | 38.50% | 0.073 | 0.595 | 0.424 | Böck |

T12 | 40 | 0.900 | 0.100 | 49.05% | 0.254 | 0.354 | 0.308 | SUP |

T12 | 163 | 0.000 | 0.605 | 38.73% | 0.197 | 0.536 | 0.404 | AD |

T12 | 200 | 0.500 | 0.700 | 33.57% | 0.119 | 0.635 | 0.457 | Böck |

T13 | 63 | 0.900 | 0.371 | 54.74% | 0.231 | 0.293 | 0.264 | SUP |

T13 | 151 | 0.643 | 0.272 | 47.92% | 0.168 | 0.434 | 0.329 | AD |

T13 | 165 | 0.819 | 0.614 | 41.67% | 0.091 | 0.551 | 0.395 | Böck |

T14 | 42 | 0.900 | 0.576 | 53.68% | 0.204 | 0.328 | 0.273 | SUP |

T14 | 120 | 0.500 | 0.100 | 38.92% | 0.156 | 0.539 | 0.397 | AD |

T14 | 200 | 0.700 | 0.100 | 21.35% | 0.059 | 0.778 | 0.552 | Böck |

T15 | 40 | 0.900 | 0.300 | 61.17% | 0.200 | 0.252 | 0.228 | SUP |

T15 | 63 | 0.900 | 0.428 | 52.67% | 0.106 | 0.421 | 0.307 | AD |

T15 | 109 | 0.900 | 0.000 | 29.95% | 0.064 | 0.687 | 0.488 | Böck |

T16 | 45 | 0.842 | 0.923 | 59.96% | 0.206 | 0.251 | 0.229 | SUP |

T16 | 101 | 0.652 | 0.762 | 47.17% | 0.116 | 0.470 | 0.342 | AD |

T16 | 154 | 0.569 | 0.621 | 35.65% | 0.086 | 0.615 | 0.439 | Böck |

T17 | 45 | 0.900 | 0.632 | 54.49% | 0.205 | 0.320 | 0.269 | SUP |

T17 | 200 | 0.104 | 0.192 | 31.68% | 0.133 | 0.637 | 0.460 | AD |

T17 | 185 | 0.603 | 0.800 | 28.57% | 0.093 | 0.691 | 0.493 | Böck |

T18 | 57 | 0.889 | 0.897 | 59.15% | 0.199 | 0.265 | 0.234 | SUP |

T18 | 116 | 0.653 | 0.370 | 49.68% | 0.172 | 0.398 | 0.307 | AD |

T18 | 160 | 0.700 | 0.300 | 39.16% | 0.094 | 0.572 | 0.410 | Böck |

T19 | 54 | 0.730 | 1.000 | 53.04% | 0.262 | 0.290 | 0.276 | SUP |

T19 | 40 | 0.900 | 0.700 | 51.35% | 0.221 | 0.343 | 0.288 | AD |

T19 | 40 | 0.601 | 0.000 | 42.72% | 0.460 | 0.223 | 0.362 | Böck |

T20 | 40 | 0.900 | 0.900 | 53.31% | 0.221 | 0.319 | 0.274 | SUP |

T20 | 200 | 0.154 | 0.961 | 34.98% | 0.131 | 0.604 | 0.437 | AD |

T20 | 200 | 0.700 | 0.500 | 24.48% | 0.067 | 0.743 | 0.528 | Böck |

T21 | 63 | 0.899 | 0.868 | 64.99% | 0.157 | 0.231 | 0.198 | SUP |

T21 | 170 | 0.627 | 0.582 | 48.55% | 0.129 | 0.447 | 0.329 | AD |

T21 | 200 | 0.813 | 0.173 | 39.54% | 0.074 | 0.579 | 0.412 | Böck |

## References

- Dudley, N.; Alexander, S. Agriculture and biodiversity: A review. Biodiversity
**2017**, 18, 45–49. [Google Scholar] [CrossRef] - Foley, J.A.; Ramankutty, N.; Brauman, K.A.; Cassidy, E.S.; Gerber, J.S.; Johnston, M.; Mueller, N.D.; O’Connell, C.; Ray, D.K.; West, P.C.; et al. Solutions for a cultivated planet. Nature
**2011**, 478, 337–342. [Google Scholar] [CrossRef] [PubMed] [Green Version] - Rey Benayas, J.M.; Bullock, J.M. Restoration of biodiversity and ecosystem services on agricultural land. Ecosystems
**2012**, 15, 883–899. [Google Scholar] [CrossRef] - Beach, R.H.; DeAngelo, B.J.; Rose, S.; Li, C.; Salas, W.; DelGrosso, S.J. Mitigation potential and costs for global agricultural greenhouse gas emissions1. Agric. Econ.
**2008**, 38, 109–115. [Google Scholar] [CrossRef] - Adams, C.R.; Eswaran, H. Global land resources in the context of food and environmental security. In Advances in Land Resources Management for the 20th Century; Soil Conservation Society of India: New Delhi, India, 2000; pp. 35–50. [Google Scholar]
- Forkuor, G.; Conrad, C.; Thiel, M.; Ullmann, T.; Zoungrana, E. integration of optical and synthetic aperture radar imagery for improving crop mapping in Northwestern Benin, West Africa. Remote Sens.
**2014**, 6, 6472–6499. [Google Scholar] [CrossRef] [Green Version] - Villa, P.; Stroppiana, D.; Fontanelli, G.; Azar, R.; Brivio, P. In-season mapping of crop type with optical and X-band SAR data: A classification tree approach using synoptic seasonal features. Remote Sens.
**2015**, 7, 12859–12886. [Google Scholar] [CrossRef] [Green Version] - Peña-Barragán, J.M.; Ngugi, M.K.; Plant, R.E.; Six, J. Object-based crop identification using multiple vegetation indices, textural features and crop phenology. Remote Sens. Environ.
**2011**, 115, 1301–1316. [Google Scholar] [CrossRef] - Blaes, X.; Vanhalle, L.; Defourny, P. Efficiency of crop identification based on optical and SAR image time series. Remote Sens. Environ.
**2005**, 96, 352–365. [Google Scholar] [CrossRef] - Castillejo-González, I.L.; López-Granados, F.; García-Ferrer, A.; Peña-Barragán, J.M.; Jurado-Expósito, M.; de la Orden, M.S.; González-Audicana, M. Object- and pixel-based analysis for mapping crops and their agro-environmental associated measures using QuickBird imagery. Comput. Electron. Agric.
**2009**, 68, 207–215. [Google Scholar] [CrossRef] - Atzberger, C. Advances in remote sensing of agriculture: Context description, existing operational monitoring systems and major information needs. Remote Sens.
**2013**, 5, 949–981. [Google Scholar] [CrossRef] [Green Version] - Blaschke, T.; Lang, S.; Lorup, E.; Strobl, J.; Zeil, P. Object-oriented image processing in an integrated GIS/remote sensing environment and perspectives for environmental applications. Environ. Inf. Plan. Polit. Public
**2000**, 2, 555–570. [Google Scholar] - Whiteside, T.G.; Boggs, G.S.; Maier, S.W. Comparing object-based and pixel-based classifications for mapping savannas. Int. J. Appl. Earth Obs. Geoinf.
**2011**, 13, 884–893. [Google Scholar] [CrossRef] - Robertson, L.D.; King, D.J. Comparison of pixel- and object-based classification in land cover change mapping. Int. J. Remote Sens.
**2011**, 32, 1505–1529. [Google Scholar] [CrossRef] - Blaschke, T. Object based image analysis for remote sensing. ISPRS J. Photogramm. Remote Sens.
**2010**, 65, 2–16. [Google Scholar] [CrossRef] [Green Version] - Georganos, S.; Lennert, M.; Grippa, T.; Vanhuysse, S.; Johnson, B.; Wolff, E. Normalization in unsupervised segmentation parameter optimization: A solution based on local regression trend analysis. Remote Sens.
**2018**, 10, 222. [Google Scholar] [CrossRef] [Green Version] - Akcay, O.; Avsar, E.O.; Inalpulat, M.; Genc, L.; Cam, A. Assessment of segmentation parameters for object-based land cover classification using color-infrared imagery. ISPRS Int. J. Geo-Inf.
**2018**, 7, 424. [Google Scholar] [CrossRef] [Green Version] - Gao, Y.; Mas, J.F.; Kerle, N.; Pacheco, J.A.N. Optimal region growing segmentation and its effect on classification accuracy. Int. J. Remote Sens.
**2011**, 32, 3747–3763. [Google Scholar] [CrossRef] - Georganos, S.; Grippa, T.; Lennert, M.; Vanhuysse, S.; Johnson, B.A.; Wolff, E. Scale Matters: Spatially Partitioned Unsupervised Segmentation Parameter Optimization for Large and Heterogeneous Satellite Images. Remote Sens.
**2018**, 10, 1440. [Google Scholar] [CrossRef] [Green Version] - Liu, D.; Xia, F. Assessing object-based classification: Advantages and limitations. Remote Sens. Lett.
**2010**, 1, 187–194. [Google Scholar] [CrossRef] - Baatz, M.; Schäpe, A. Multiresolution segmentation: An optimization approach for high quality multi-scale image segmentation. In Proceedings of the Angewandte Geographische Informations-Verarbeitung XII, Karlsruhe, Germany, 30 June 2000; Strobl, J., Blaschke, T., Griesebner, G., Eds.; Wichmann Verlag: Karlsruhe, Germany, 2000; pp. 12–23. [Google Scholar]
- Benz, U.C.; Hofmann, P.; Willhauck, G.; Lingenfelder, I.; Heynen, M. Multi-resolution, object-oriented fuzzy analysis of remote sensing data for GIS-ready information. ISPRS J. Photogramm. Remote Sens.
**2004**, 58, 239–258. [Google Scholar] [CrossRef] - Ma, L.; Li, M.; Ma, X.; Cheng, L.; Du, P.; Liu, Y. A review of supervised object-based land-cover image classification. ISPRS J. Photogramm. Remote Sens.
**2017**, 130, 277–293. [Google Scholar] [CrossRef] - Trimble Germany GmbH. eCognition Developer 9.5.0 Reference Book; Trimble Germany GmbH: Munich, Germany, 2019. [Google Scholar]
- Marpu, P.R.; Neubert, M.; Herold, H.; Niemeyer, I. Enhanced evaluation of image segmentation results. J. Spat. Sci.
**2010**, 55, 55–68. [Google Scholar] [CrossRef] - Neubert, M.; Herold, H.; Meinel, G. Assessing image segmentation quality—Concepts, methods and application. In Object-Based Image Analysis: Spatial Concepts for Knowledge-Driven Remote Sensing Applications; Blaschke, T., Lang, S., Hay, G.J., Eds.; Lecture Notes in Geoinformation and Cartography; Springer: Berlin/Heidelberg, Germany, 2008; pp. 769–784. ISBN 978-3-540-77058-9. [Google Scholar]
- Neubert, M.; Meinel, G. Evaluation of segmentation programs for high resolution remote sensing applications. In Proceedings of the Joint ISPRS/EARSeL Workshop “High Resolution Mapping from Space 2003”, Hannover, Germany, 8 October 2003. [Google Scholar]
- Zhang, Y.-J. A survey on evaluation methods for image segmentation. Pattern Recognit.
**1996**, 29, 1335–1346. [Google Scholar] [CrossRef] [Green Version] - Zhang, H.; Fritts, J.E.; Goldman, S.A. Image segmentation evaluation: A survey of unsupervised methods. Comput. Vis. Image Underst.
**2008**, 110, 260–280. [Google Scholar] [CrossRef] [Green Version] - Chabrier, S.; Emile, B.; Rosenberger, C.; Laurent, H. Unsupervised Performance Evaluation of Image Segmentation. EURASIP J. Adv. Signal Process.
**2006**, 2006, 096306. [Google Scholar] [CrossRef] [Green Version] - Drăguţ, L.; Csillik, O.; Eisank, C.; Tiede, D. Automated parameterisation for multi-scale image segmentation on multiple layers. ISPRS J. Photogramm. Remote Sens.
**2014**, 88, 119–127. [Google Scholar] [CrossRef] [Green Version] - Drǎguţ, L.; Tiede, D.; Levick, S.R. ESP: A tool to estimate scale parameter for multiresolution image segmentation of remotely sensed data. Int. J. Geogr. Inf. Sci.
**2010**, 24, 859–871. [Google Scholar] [CrossRef] - Espindola, G.M.; Camara, G.; Reis, I.A.; Bins, L.S.; Monteiro, A.M. Parameter selection for region-growing image segmentation algorithms using spatial autocorrelation. Int. J. Remote Sens.
**2006**, 27, 3035–3040. [Google Scholar] [CrossRef] - Woodcock, C.E.; Strahler, A.H. The factor of scale in remote sensing. Remote Sens. Environ.
**1987**, 21, 311–332. [Google Scholar] [CrossRef] - Moran, P.A.P. Notes on continuous stochastic phenomena. Biometrika
**1950**, 37, 17–23. [Google Scholar] [CrossRef] - Grybas, H.; Melendy, L.; Congalton, R.G. A comparison of unsupervised segmentation parameter optimization approaches using moderate- and high-resolution imagery. GISci. Remote Sens.
**2017**, 54, 515–533. [Google Scholar] [CrossRef] - Johnson, B.; Xie, Z. Unsupervised image segmentation evaluation and refinement using a multi-scale approach. ISPRS J. Photogramm. Remote Sens.
**2011**, 66, 473–483. [Google Scholar] [CrossRef] - Johnson, B.A.; Bragais, M.; Endo, I.; Magcale-Macandog, D.B.; Macandog, P.B.M. Image segmentation parameter optimization considering within- and between-segment heterogeneity at multiple scale levels: Test case for mapping residential areas using landsat imagery. ISPRS Int. J. Geo-Inf.
**2015**, 4, 2292–2305. [Google Scholar] [CrossRef] [Green Version] - Yang, L.; Mansaray, L.R.; Huang, J.; Wang, L. Optimal Segmentation Scale Parameter, Feature Subset and Classification Algorithm for Geographic Object-Based Crop Recognition Using Multisource Satellite Imagery. Remote Sens.
**2019**, 11, 514. [Google Scholar] [CrossRef] [Green Version] - Kim, M.; Madden, M.; Warner, T. Estimation of optimal image object size for the segmentation of forest stands with multispectral IKONOS imagery. In Object-Based Image Analysis; Blaschke, T., Lang, S., Hay, G.J., Eds.; Lecture Notes in Geoinformation and Cartography; Springer: Berlin/Heidelberg, Germany, 2008; pp. 291–307. ISBN 978-3-540-77057-2. [Google Scholar]
- Martha, T.R.; Kerle, N.; van Westen, C.J.; Jetten, V.; Kumar, K.V. Segment optimization and data-driven thresholding for knowledge-based landslide detection by object-based image analysis. IEEE Trans. Geosci. Remote Sens.
**2011**, 49, 4928–4943. [Google Scholar] [CrossRef] - Chen, J.; Deng, M.; Mei, X.; Chen, T.; Shao, Q.; Hong, L. Optimal segmentation of a high-resolution remote-sensing image guided by area and boundary. Int. J. Remote Sens.
**2014**, 35, 6914–6939. [Google Scholar] [CrossRef] - Böck, S.; Immitzer, M.; Atzberger, C. On the objectivity of the objective function—Problems with unsupervised segmentation evaluation based on global score and a possible remedy. Remote Sens.
**2017**, 9, 769. [Google Scholar] [CrossRef] [Green Version] - Kim, M.; Warner, T.A.; Madden, M.; Atkinson, D.S. Multi-scale GEOBIA with very high spatial resolution digital aerial imagery: Scale, texture and image objects. Int. J. Remote Sens.
**2011**, 32, 2825–2850. [Google Scholar] [CrossRef] - Johnson, B.; Xie, Z. Classifying a high resolution image of an urban area using super-object information. ISPRS J. Photogramm. Remote Sens.
**2013**, 83, 40–49. [Google Scholar] [CrossRef] - Taşdemir, K.; Wirnhardt, C. Neural network-based clustering for agriculture management. EURASIP J. Adv. Signal Process.
**2012**, 2012. [Google Scholar] [CrossRef] - Tetteh, G.O.; Gocht, A.; Conrad, C. Optimal parameters for delineating agricultural parcels from satellite images based on supervised Bayesian optimization. Comput. Electron. Agric.
**2020**, 178, 105696. [Google Scholar] [CrossRef] - Main-Knorn, M.; Pflug, B.; Louis, J.; Debaecker, V.; Müller-Wilm, U.; Gascon, F. Sen2Cor for Sentinel-2. In Proceedings of the Image and Signal Processing for Remote Sensing XXIII, Warsaw, Poland, 11–13 September 2017; Bruzzone, L., Bovolo, F., Benediktsson, J.A., Eds.; SPIE: Warsaw, Poland, 2017; p. 3. [Google Scholar]
- Rousseeuw, P.J. Silhouettes: A graphical aid to the interpretation and validation of cluster analysis. J. Comput. Appl. Math.
**1987**, 20, 53–65. [Google Scholar] [CrossRef] [Green Version] - Rasmussen, C.E.; Williams, C.K.I. Gaussian Processes for Machine Learning; Adaptive Computation and Machine Learning; MIT Press: Cambridge, MA, USA, 2006; ISBN 978-0-262-18253-9. [Google Scholar]
- Jones, D.R.; Schonlau, M.; Welch, W.J. Efficient global optimization of expensive black-box functions. J. Glob. Optim.
**1998**, 13, 455–492. [Google Scholar] [CrossRef] - Dewancker, I.; McCourt, M.; Clark, S. Bayesian Optimization Primer. Available online: https://app.sigopt.com/static/pdf/SigOpt_Bayesian_Optimization_Primer.pdf (accessed on 4 March 2020).
- Shahriari, B.; Swersky, K.; Wang, Z.; Adams, R.P.; de Freitas, N. Taking the human out of the loop: A review of Bayesian optimization. Proc. IEEE
**2016**, 104, 148–175. [Google Scholar] [CrossRef] [Green Version] - Brochu, E.; Cora, V.M.; de Freitas, N. A tutorial on Bayesian optimization of expensive cost functions, with application to active user modeling and hierarchical reinforcement learning. arXiv
**2010**, arXiv:10122599. [Google Scholar] - Frazier, P.I. A tutorial on Bayesian optimization. arXiv
**2018**, arXiv:180702811. [Google Scholar] - Weidner, U. Contribution to the assessment of segmentation quality for remote sensing applications. Int. Arch. Photogramm. Remote Sens. Spat. Inf. Sci.
**2008**, 37, 479–484. [Google Scholar] - Clinton, N.; Holt, A.; Scarborough, J.; Yan, L.; Gong, P. Accuracy assessment measures for object-based image segmentation goodness. Photogramm. Eng. Remote Sens.
**2010**, 76, 289–299. [Google Scholar] [CrossRef] - Liu, Y.; Bian, L.; Meng, Y.; Wang, H.; Zhang, S.; Yang, Y.; Shao, X.; Wang, B. Discrepancy measures for selecting optimal combination of parameter values in object-based image analysis. ISPRS J. Photogramm. Remote Sens.
**2012**, 68, 144–156. [Google Scholar] [CrossRef] - Belgiu, M.; Drǎguţ, L. Comparing supervised and unsupervised multiresolution segmentation approaches for extracting buildings from very high resolution imagery. ISPRS J. Photogramm. Remote Sens.
**2014**, 96, 67–75. [Google Scholar] [CrossRef] [Green Version]

**Figure 1.**The study sites (tiles) overlaid on a mosaic of cloud-free and non-masked Sentinel-2 images captured in May 2018. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 2.**The simplified workflow we used in this study. Böck refers to the unsupervised segmentation evaluation metric proposed by Böck et al. [43], and absolute difference (AD) is the modified version we proposed in this study.

**Figure 3.**Simulated reference and segmentation data. The reference dataset is represented by (

**a**). Three different corresponding segmentation results are represented by (

**b**–

**d**), respectively. Each row in each dataset represents a segment; hence, there are four segments in all.

**Figure 4.**The quality rate (QR) measure computed for each optimal segmentation result identified by the AD and Böck metrics based on the default optimization (shape = 0.1, compactness = 0.5).

**Figure 5.**Examples of segments identified as optimal at T1 using the default shape and compactness parameters. (

**a**) An example based on the optimal segmentation identified by the Böck metric showing massive under-segmentation and (

**b**) based on the AD metric, which shows a better delineation of the agricultural parcels with lower under-segmentation compared to Böck. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 6.**The normalized average area-weighted variance (nWV), Moran’s I (MI), normalized Moran’s I (nMI), and global score (GS) computed for each scale at T1 based on (

**a**) the Böck metric and (

**b**) the AD metric.

**Figure 7.**The normalized average area-weighted variance (nWV), Moran’s I (MI), normalized Moran’s I (nMI), and global score (GS) computed for each scale at T3 based on (

**a**) the Böck metric and (

**b**) the AD metric.

**Figure 8.**The quality rate (QR) measure computed for the unsupervised Bayesian optimization approaches based on the Böck and AD metrics, and the supervised Bayesian optimization (SUP) approach that was used to maximize the QR measure by Tetteh et al. [47].

**Figure 9.**The outcome of the three Bayesian optimization approaches at T1. The Sentinel-2 image is shown by (

**a**). The optimal segments as identified by (

**b**) the Böck metric, (

**c**) the supervised Bayesian optimization approach, and (

**d**) the AD metric are symbolized by their respective QR measures. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 10.**The outcome of the three Bayesian optimization approaches at T2. The Sentinel-2 image is shown by (

**a**). The optimal segments as identified by (

**b**) the Böck metric, (

**c**) the supervised Bayesian optimization approach, and (

**d**) the AD metric are symbolized by their respective QR measures. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 11.**The outcome of the three Bayesian optimization approaches at T17. The Sentinel-2 image is shown by (

**a**). The optimal segments as identified by (

**b**) the Böck metric, (

**c**) the supervised Bayesian optimization approach, and (

**d**) the AD metric are symbolized by their respective QR measures. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 12.**An example of segments created at T1 using the unsupervised Bayesian optimization approach based on (

**a**) the Böck metric and (

**b**) the AD metric. (

**c**) Segments generated by the supervised Bayesian optimization approach (SUP) based on the QR metric. The coordinates are in UTM Zone 32N (EPSG:32632).

**Figure 13.**Correlation between the number of land use (LU) types and the difference in QR between the supervised benchmark results and the unsupervised Bayesian optimization approaches based on (

**a**) the Böck metric and (

**b**) the AD metric.

**Table 1.**The Moran’s I (MI), normalized MI (nMI), normalized weighted variance (nWV), and global score (GS) of the Böck metric computed for the simulated data at Figure 3. The bold-faced text within the body of the table is the optimal result.

Identifier | MI | nMI | nWV | GS (Böck) |
---|---|---|---|---|

Figure 3b | 0.400 | 0.700 | 0.375 | 1.075 |

Figure 3c | −0.018 | 0.491 | 0.698 | 1.189 |

Figure 3d | −0.667 | 0.167 | 0.875 | 1.042 |

**Table 2.**The MI, nWV, and GS of the AD metric computed for the simulated data in Figure 3. The bold-faced text within the body of the table is the optimal result.

Measure | Formula | Range | Source |
---|---|---|---|

Quality rate (QR) | $\frac{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)\ast \frac{Area\left({X}_{i}{{\displaystyle \cap}}^{}{Y}_{i}\right)}{Area\left({X}_{i}{{\displaystyle \cup}}^{}{Y}_{i}\right)}}{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)}$ | 0 (worst) to 1 (perfect) segmentation | [56] |

Over-segmentation (OR) | $1-\frac{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)\ast \frac{Area\left({X}_{i}{{\displaystyle \cap}}^{}{Y}_{i}\right)}{Area\left({X}_{i}\right)}}{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)}$ | 0 (perfect) to 1 (worst) segmentation | [57] |

Under-segmentation (UR) | $1-\frac{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)\ast \frac{Area\left({X}_{i}{{\displaystyle \cap}}^{}{Y}_{i}\right)}{Area\left({Y}_{i}\right)}}{{{\displaystyle \sum}}_{i=1}^{n}Area\left({Y}_{i}\right)}$ | 0 (perfect) to 1 (worst) segmentation | [57] |

Root mean square (RMS) | $\sqrt{\frac{O{R}^{2}+U{R}^{2}}{2}}$ | 0 (perfect) to 1 (worst) segmentation | [56] |

© 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**

Tetteh, G.O.; Gocht, A.; Schwieder, M.; Erasmi, S.; Conrad, C.
Unsupervised Parameterization for Optimal Segmentation of Agricultural Parcels from Satellite Images in Different Agricultural Landscapes. *Remote Sens.* **2020**, *12*, 3096.
https://doi.org/10.3390/rs12183096

**AMA Style**

Tetteh GO, Gocht A, Schwieder M, Erasmi S, Conrad C.
Unsupervised Parameterization for Optimal Segmentation of Agricultural Parcels from Satellite Images in Different Agricultural Landscapes. *Remote Sensing*. 2020; 12(18):3096.
https://doi.org/10.3390/rs12183096

**Chicago/Turabian Style**

Tetteh, Gideon Okpoti, Alexander Gocht, Marcel Schwieder, Stefan Erasmi, and Christopher Conrad.
2020. "Unsupervised Parameterization for Optimal Segmentation of Agricultural Parcels from Satellite Images in Different Agricultural Landscapes" *Remote Sensing* 12, no. 18: 3096.
https://doi.org/10.3390/rs12183096