Evaluation of Three Different Machine Learning Methods for Object-Based Artificial Terrace Mapping—A Case Study of the Loess Plateau, China

Ding, Hu; Na, Jiaming; Jiang, Shangjing; Zhu, Jie; Li, Kai; Fu, Yingchun; Li, Fayuan

doi:10.3390/rs13051021

Cited by 22 publications

(14 citation statements)

References 58 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In order to reduce overfitting, we repeated these steps 10 times, for both training and accuracy assessment for each classifier (Figure 5). We then employed the F1 Score (FS) to assess the mapping accuracy [39]. The mean value (MV) and standard deviation (SD) of the FS and kappa coefficient (KA) were used to assess the mapping accuracy and stability of the classifiers (Figure 5).…”

Section: Validation and Comparisonmentioning

confidence: 99%

Integrating Phenological and Geographical Information with Artificial Intelligence Algorithm to Map Rubber Plantations in Xishuangbanna

Yang

Zhai

2021

Remote Sensing

View full text Add to dashboard Cite

Most natural rubber trees (Hevea brasiliensis) are grown on plantations, making rubber an important industrial crop. Rubber plantations are also an important source of household income for over 20 million people. The accurate mapping of rubber plantations is important for both local governments and the global market. Remote sensing has been a widely used approach for mapping rubber plantations, typically using optical remote sensing data obtained at the regional scale. Improving the efficiency and accuracy of rubber plantation maps has become a research hotspot in rubber-related literature. To improve the classification efficiency, researchers have combined the phenology, geography, and texture of rubber trees with spectral information. Among these, there are three main classifiers: maximum likelihood, QUEST decision tree, and random forest methods. However, until now, no comparative studies have been conducted for the above three classifiers. Therefore, in this study, we evaluated the mapping accuracy based on these three classifiers, using four kinds of data input: Landsat spectral information, phenology–Landsat spectral information, topography–Landsat spectral information, and phenology–topography–Landsat spectral information. We found that the random forest method had the highest mapping accuracy when compared with the maximum likelihood and QUEST decision tree methods. We also found that adding either phenology or topography could improve the mapping accuracy for rubber plantations. When either phenology or topography were added as parameters within the random forest method, the kappa coefficient increased by 5.5% and 6.2%, respectively, compared to the kappa coefficient for the baseline Landsat spectral band data input. The highest accuracy was obtained from the addition of both phenology–topography–Landsat spectral bands to the random forest method, achieving a kappa coefficient of 97%. We therefore mapped rubber plantations in Xishuangbanna using the random forest method, with the addition of phenology and topography information from 1990–2020. Our results demonstrated the usefulness of integrating phenology and topography for mapping rubber plantations. The machine learning approach showed great potential for accurate regional mapping, particularly by incorporating plant habitat and ecological information. We found that during 1990–2020, the total area of rubber plantations had expanded to over three times their former area, while natural forests had lost 17.2% of their former area.

show abstract

Section: Validation and Comparisonmentioning

confidence: 99%

Integrating Phenological and Geographical Information with Artificial Intelligence Algorithm to Map Rubber Plantations in Xishuangbanna

Yang

Zhai

2021

Remote Sensing

View full text Add to dashboard Cite

show abstract

“…An overview of these features and the detailed description for each feature are listed in Table 1. As spectral and topographic features are widely used in the OBIA community [56,66,70,71], this information was firstly considered. For the spectral features, the R, G, and B bands from the images and a vegetation index based on the above-mentioned three features called EXG were considered.…”

Section: Image Segmentationmentioning

confidence: 99%

“…Therefore, the proposed method has application potential in the damage extraction of horizontal terraces covered grasslands or bare lands. When the study area includes terracing area and non-terracing area, a two-step scheme can be adopted: firstly, extracting the terrace area via the terrace extraction method [52,66], and secondly, using the proposed method for extracting damages in the terrace area.…”

Section: Potential Applicationmentioning

confidence: 99%

“…Moreover, owing to the use of high-resolution data with high spectral variety between pixels, which often results in oversampling, the approach of OBIA of clustering pixels into objects is more effective than that of the pixel-based method [57] in processing high-resolution images requiring terrace damage extraction. The recent studies have also proven that OBIA and UAV data can be successfully applied in many fields, such as building safety [58][59][60][61], crop production [62,63], landslide movement [64,65], and artificial terrace mapping [52,66].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

OBIA-Based Extraction of Artificial Terrace Damages in the Loess Plateau of China from UAV Photogrammetry

Fang

Zhu

et al. 2021

IJGI

Self Cite

View full text Add to dashboard Cite

Terraces, which are typical artificial landforms found around world, are of great importance for agricultural production and soil and water conservation. However, due to the lack of maintenance, terrace damages often occur and affect the local flow process, which will influence soil erosion. Automatic high-accuracy mapping of terrace damages is the basis of monitoring and related studies. Researchers have achieved artificial terrace damage mapping mainly via manual field investigation, but an automatic method is still lacking. In this study, given the success of high-resolution unmanned aerial vehicle (UAV) photogrammetry and object-based image analysis (OBIA) for image processing tasks, an integrated framework based on OBIA and UAV photogrammetry is proposed for terrace damage mapping. The Pujiawa terrace in the Loess Plateau of China was selected as the study area. Firstly, the segmentation process was optimised by considering the spectral features and the terrains and corresponding textures obtained from high-resolution images and digital surface models. The feature selection was implemented via correlation analysis, and the optimised segmentation parameter was achieved using the estimation of scale parameter algorithm. Then, a supervised k-nearest neighbourhood classifier was used to identify the terrace damages in the segmented objects, and additional geometric features at the object level were considered for classification. The comparison with the ground truth, as delineated by the image and field survey, showed that proposed classification can be adequately performed. The F-measures of extraction on three terrace damages were 92.07% (terrace sinkhole), 81.95% (ridge sinkhole), and 85.17% (collapse), and the Kappa coefficient was 85.34%. Finally, the potential application and spatial distribution of the terrace damages in this study were determined. We believe that this work can provide a credible framework for mapping terrace damages in the Loess Plateau of China.

show abstract

“…Currently, the research on loess landforms mainly focuses on four aspects, namely (1) classification methods of loess landforms [12][13][14], (2) extraction and analysis of terrain derivatives of loess landforms [15,16], (3) analysis on development and evolution process of loess landforms [5,7,17], and (4) construction of quantitative indices of loess landforms [18][19][20]. More specifically, regarding the classification methods of loess landforms, the current classification methods can be divided into two categories: pixel-based and object-based landform classification [13,21]. Despite the long-term dominance of pixel-based image analysis methods in remote sensing image processing, object-based image analysis is becoming increasingly popular [22].…”

Section: Introductionmentioning

confidence: 99%

Quantification of Loess Landforms from Three-Dimensional Landscape Pattern Perspective by Using DEMs

Wei

et al. 2021

IJGI

View full text Add to dashboard Cite

Quantitative analysis of the differences and the exploration of the evolution models of different loess landform types are greatly important to the in-depth understanding of the evolution process and mechanism of the loess landforms. In this research, several typical loess landform areas in the Chinese Loess Plateau were selected, and the object-oriented image analysis (OBIA) method was employed to identify the basic loess landform types. Three-dimensional (3D) landscape pattern indices were introduced on this foundation to measure the morphological and structural features of individual loess landform objects in more detail. Compared with the traditional two-dimensional (2D) landscape pattern indices, the indices consider the topographic features, thereby providing more vertical topographic information. Furthermore, the evolution modes between different loess landform types were discussed. Results show that the OBIA method achieved satisfying classification results with an overall accuracy of 88.12%. There are evident differences in quantitative morphological indicators among loess landform types, especially in indicators such as total length of edge, mean patch size, landscape shape index, and edge dimension index. Meanwhile, significant differences are also found in the combination of loess landform types corresponding to different landform development stages. The degree of surface erosion became increasingly significant as loess landforms developed, loess tableland area rapidly reduced or even vanished, and the dominant loess landform types changed to loess ridge and loess hill. Hence, in the reconstruction and management of the Loess Plateau, the loess tableland should be the key protected loess landform type. These preliminary results are helpful to further understand the development process of loess landforms and provide a certain reference for regional soil and water conservation.

show abstract

Evaluation of Three Different Machine Learning Methods for Object-Based Artificial Terrace Mapping—A Case Study of the Loess Plateau, China

Cited by 22 publications

References 58 publications

Integrating Phenological and Geographical Information with Artificial Intelligence Algorithm to Map Rubber Plantations in Xishuangbanna

Integrating Phenological and Geographical Information with Artificial Intelligence Algorithm to Map Rubber Plantations in Xishuangbanna

OBIA-Based Extraction of Artificial Terrace Damages in the Loess Plateau of China from UAV Photogrammetry

Quantification of Loess Landforms from Three-Dimensional Landscape Pattern Perspective by Using DEMs

Contact Info

Product

Resources

About