Multitemporal Crop Type Classification Using Conditional Random Fields and Rapideye Data

Hoberg, Thorsten; Müller, Sönke

doi:10.5194/isprsarchives-xxxviii-4-w19-115-2011

Cited by 7 publications

(5 citation statements)

References 27 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…International Journal of Remote Sensing 7249 improve the accuracy of vegetation classification (Hoberg and Müller 2011;Tapsall, Pavel, and Kadim 2010). In this research, the multi-temporal REVI was used as an indicator for the seasonal changes of vegetation.…”

Section: Pixel-based Image Analysismentioning

confidence: 99%

Extraction of small biotopes and ecotones from multi-temporal RapidEye data and a high-resolution normalized digital surface model

Hou

Walz

2014

International Journal of Remote Sensing

View full text Add to dashboard Cite

The conservation of habitats and habitat complexes in diverse landscapes is increasingly recognized as a crucial factor in sustaining biodiversity and ecosystems. For the successful management of landscapes, habitat monitoring is necessary, but often small biotopes, e.g. scattered trees, copses, tree rows, hedges, and the transition zones between ecosystems, are ignored. This is important as such small biotopes are recognized as keystone elements in landscape structure for habitat networks. Furthermore, the transition zones between different habitats, often called ecotones, are dynamic and play several functional roles in landscape ecology. This article presents an approach for the extraction of small biotopes and ecotones combining object-based and pixel-based image analysis. Both high-resolution digital elevation data from airborne laser scanning and multi-temporal RapidEye remote-sensing data were used to automatically detect landscape elements and landscape patterns. First, multi-temporal RapidEye images were used to classify the main land-use classes using object-based image analysis. In the second step, a high-resolution digital surface model was integrated with the main classes, and small biotopes and ecotones were delineated by means of pixel-based image analysis. Classification accuracy for main land-use classes is above 92%, and a visual assessment using aerial image and onsite investigation show that the identification results for small biotopes well match reality. The results show the effectiveness of the classification strategy developed and the potential for incorporating the detailed surface mapping in heterogeneous vegetated areas.

show abstract

Section: Pixel-based Image Analysismentioning

confidence: 99%

Extraction of small biotopes and ecotones from multi-temporal RapidEye data and a high-resolution normalized digital surface model

Hou

Walz

2014

International Journal of Remote Sensing

View full text Add to dashboard Cite

show abstract

“…However, in highly fractured landscapes with heterogeneous cropping patterns of smaller fields, images with higher spatial resolution are needed [11], [12]. BlackBridge's RapidEye constellation consisting of 5 identical satellites provides high spatial resolution images (5 m) with a theoretical repetition rate of 4-5 days, which makes it highly appropriate for accurate crop type mapping.…”

Section: Introductionmentioning

confidence: 99%

Crop identification by means of seasonal statistics of RapidEye time series

Zillmann¹,

Weichelt²

2014

2014 the Third International Conference on Agro-Geoinformatics

View full text Add to dashboard Cite

Crop classification greatly benefits from the analysis of multi-temporal Earth Observation (EO) data within a growing season utilizing the distinct phenological behavior of each crop. RapidEye's high repetition rate increases the chances of providing sufficient high resolution image time series offering new ways of classifying crops. This study introduces a supervised decision tree (DT) classification approach using image objects in combination with seasonal statistics of various vegetation indices (VI) for crop identification. The aim of this study is, first, to show the potential of VI seasonal statistics for crop identification, and secondly, to evaluate the relative contribution of each variable to the overall classification accuracy. The results presented in this paper correspond to an area of 625 km² in Saxony-Anhalt, Germany. The cultivated landscape is characterized by large agricultural fields, with winter wheat, canola, corn and winter barley as the main crops. Crop identification accuracies were assessed on the basis of reference fields and the importance of each employed variable is assessed by rule set analysis. The classification accuracy for the test area demonstrates that the proposed approach of multi-temporal image analysis provides spatially detailed and thematically accurate information on the crop type distribution.

show abstract

“…Mono-temporal crop classification adopting spatial context also exist (Ozdarici-Ok et al, 2015;Roscher et al, 2010). Efforts to use both spatial and temporal context to classify crops from optical images is done in (Hoberg and Müller, 2011). The study models temporal context via a global transition matrix determined from training data.…”

Section: Introductionmentioning

confidence: 99%

Crop Type Mapping From a Sequence of Terrasar-X Images With Dynamic Conditional Random Fields

Kenduiywo

Bargiel

Soergel

2016

ISPRS Ann. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

Crop phenology is dynamic as it changes with times of the year. Such biophysical processes also look spectrally different to remote sensing satellites. Some crops may depict similar spectral properties if their phenology coincide, but differ later when their phenology diverge. Thus, conventional approaches that select only images from phenological stages where crops are distinguishable for classification, have low discrimination. In contrast, stacking images within a cropping season limits discrimination to a single feature space that can suffer from overlapping classes. Since crop backscatter varies with time, it can aid discrimination. Therefore, our main objective is to develop a crop sequence classification method using multitemporal TerraSAR-X images. We adopt first order markov assumption in undirected temporal graph sequence. This property is exploited to implement Dynamic Conditional Random Fields (DCRFs). Our DCRFs model has a repeated structure of temporally connected Conditional Random Fields (CRFs). Each node in the sequence is connected to its predecessor via conditional probability matrix. The matrix is computed using posterior class probabilities from association potential. This way, there is a mutual temporal exchange of phenological information observed in TerraSAR-X images. When compared to independent epoch classification, the designed DCRF model improved crop discrimination at each epoch in the sequence. However, government, insurers, agricultural market traders and other stakeholders are interested in the quantity of a certain crop in a season. Therefore, we further develop a DCRF ensemble classifier. The ensemble produces an optimal crop map by maximizing over posterior class probabilities selected from the sequence based on maximum F1-score and weighted by correctness. Our ensemble technique is compared to standard approach of stacking all images as bands for classification using Maximum Likelihood Classifier (MLC) and standard CRFs. It outperforms MLC and CRFs by 7.70% and 6.42% in overall accuracy, respectively.

show abstract

Multitemporal Crop Type Classification Using Conditional Random Fields and Rapideye Data

Cited by 7 publications

References 27 publications

Extraction of small biotopes and ecotones from multi-temporal RapidEye data and a high-resolution normalized digital surface model

Extraction of small biotopes and ecotones from multi-temporal RapidEye data and a high-resolution normalized digital surface model

Crop identification by means of seasonal statistics of RapidEye time series

Crop Type Mapping From a Sequence of Terrasar-X Images With Dynamic Conditional Random Fields

Contact Info

Product

Resources

About