Assessing Different Feature Sets’ Effects on Land Cover Classification in Complex Surface-Mined Landscapes by ZiYuan-3 Satellite Imagery

Chen, Weitao; Li, Xianju; He, Honglin; Wang, Lizhe

doi:10.3390/rs10010023

Cited by 44 publications

(48 citation statements)

References 38 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A 109.4 km 2 area in Wuhan City, China that was used by previous studies [12,14] was further tested in this study (Figure 1). The study area covers a CMAL with a heterogeneous terrain and agricultural development in different phenological periods [12].…”

Section: Study Area and Data Sourcesmentioning

confidence: 99%

“…LCC of complex open pit mining and agricultural development landscapes (CMALs) was studied previously [12][13][14]. In particular, Li et al [12] conducted fine classification of three open pit mining and logistic regression (LR) to obtain a DBN-LR algorithm.…”

Section: Introductionmentioning

confidence: 99%

“…However, the FLCC of CMALs has not been investigated. This is significant for the scientific management of mines and agricultural development.Three significant landscape-specific characteristics seriously limit the accuracy of FLCC of CMALs [12][13][14]: significant three-dimensional terrain [15,16]; strong spatio-temporal variability [17,18]; and spectral-spatial homogeneity. Methods to reduce these characteristics were identified.…”

mentioning

confidence: 99%

“…Methods to reduce these characteristics were identified. These include the extraction of effective features [12,13,[19][20][21] and feature sets [13,14,22] with feature reduction (FR) [12] and feature selection (FS) [12][13][14]23] and the construction of powerful classification models [13].Many low-level handcrafted features were proven to be effective for the LCC of CMALs [12-14] (e.g., texture and filter features and topographic information). Moreover, fusion of these features is necessary and one of the most effective approaches for this purpose.…”

mentioning

confidence: 99%

See 3 more Smart Citations

Multimodal and Multi-Model Deep Fusion for Fine Classification of Regional Complex Landscape Areas Using ZiYuan-3 Imagery

Tang

Chen

et al. 2019

Remote Sensing

Self Cite

View full text Add to dashboard Cite

Land cover classification (LCC) of complex landscapes is attractive to the remote sensing community but poses great challenges. In complex open pit mining and agricultural development landscapes (CMALs), the landscape-specific characteristics limit the accuracy of LCC. The combination of traditional feature engineering and machine learning algorithms (MLAs) is not sufficient for LCC in CMALs. Deep belief network (DBN) methods achieved success in some remote sensing applications because of their excellent unsupervised learning ability in feature extraction. The usability of DBN has not been investigated in terms of LCC of complex landscapes and integrating multimodal inputs. A novel multimodal and multi-model deep fusion strategy based on DBN was developed and tested for fine LCC (FLCC) of CMALs in a 109.4 km 2 area of Wuhan City, China. First, low-level and multimodal spectral–spatial and topographic features derived from ZiYuan-3 imagery were extracted and fused. The features were then input into a DBN for deep feature learning. The developed features were fed to random forest and support vector machine (SVM) algorithms for classification. Experiments were conducted that compared the deep features with the softmax function and low-level features with MLAs. Five groups of training, validation, and test sets were performed with some spatial auto-correlations. A spatially independent test set and generalized McNemar tests were also employed to assess the accuracy. The fused model of DBN-SVM achieved overall accuracies (OAs) of 94.74%± 0.35% and 81.14% in FLCC and LCC, respectively, which significantly outperformed almost all other models. From this model, only three of the twenty land covers achieved OAs below 90%. In general, the developed model can contribute to FLCC and LCC in CMALs, and more deep learning algorithm-based models should be investigated in future for the application of FLCC and LCC in complex landscapes.

show abstract

Section: Study Area and Data Sourcesmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

mentioning

confidence: 99%

mentioning

confidence: 99%

See 2 more Smart Citations

Multimodal and Multi-Model Deep Fusion for Fine Classification of Regional Complex Landscape Areas Using ZiYuan-3 Imagery

Tang

Chen

et al. 2019

Remote Sensing

Self Cite

View full text Add to dashboard Cite

show abstract

“…Texture [1,2] is a core property of object appearance in natural scenes, ranging from large-scale samples to microscopic ones. It is also active visual information that is used to describe and recognize objects in the real environment.…”

Section: Introductionmentioning

confidence: 99%

A Texture Classification Approach Based on the Integrated Optimization for Parameters and Features of Gabor Filter via Hybrid Ant Lion Optimizer

et al. 2019

View full text Add to dashboard Cite

Texture classification is an important topic for many applications in machine vision and image analysis, and Gabor filter is considered one of the most efficient tools for analyzing texture features at multiple orientations and scales. However, the parameter settings of each filter are crucial for obtaining accurate results, and they may not be adaptable to different kinds of texture features. Moreover, there is redundant information included in the process of texture feature extraction that contributes little to the classification. In this paper, a new texture classification technique is detailed. The approach is based on the integrated optimization of the parameters and features of Gabor filter, and obtaining satisfactory parameters and the best feature subset is viewed as a combinatorial optimization problem that can be solved by maximizing the objective function using hybrid ant lion optimizer (HALO). Experimental results, particularly fitness values, demonstrate that HALO is more effective than the other algorithms discussed in this paper, and the optimal parameters and features of Gabor filter are balanced between efficiency and accuracy. The method is feasible, reasonable, and can be utilized for practical applications of texture classification.

show abstract

Detecting northern peatland vegetation patterns at ultra‐high spatial resolution

Räsänen

Aurela

Juutinen

et al. 2019

Remote Sens Ecol Conserv

View full text Add to dashboard Cite

Within northern peatlands, landscape elements such as vegetation and topography are spatially heterogenic from ultra-high (centimeter level) to coarse scale. In addition to within-site spatial heterogeneity, there is evident between-site heterogeneity, but there is a lack of studies assessing whether different combinations of remotely sensed features and mapping approaches are needed in different types of landscapes. We evaluated the value of different mapping methods and remote sensing datasets and analyzed the kinds of differences present in vegetation patterns and their mappability between three northern boreal peatland landscapes in northern Finland. We utilized field-inventoried vegetation plots together with spectral, textural, topography and vegetation height remote sensing data from 0.02-to 3-m pixel size. Remote sensing data included truecolor unmanned aerial vehicle images, aerial images with four spectral bands, aerial lidar data and multiple PlanetScope satellite images. We used random forest regressions for tracking plant functional type (PFT) coverage, non-metric multidimensional scaling ordination axes and fuzzy k-medoid plant community clusters. PFT regressions had variable performance for different study sites (R 2 À0.03 to 0.69). Spatial patterns of some spectrally or structurally distinctive PFTs could be predicted relatively well. The first ordination axis represented wetness gradient and was well predicted using remotely sensed data (R 2 0.64 to 0.82), but the other three axes had a less straightforward explanation and lower mapping performance (R 2 À0.09 to 0.53). Plant community clusters were predicted most accurately in the sites with clear string-flark topography but less accurately in the flatter site (R 2 0.16-0.82). The most important remote sensing features differed between dependent variables and study sites: different topographic, spectral and textural features; and coarse-scale and fine-scale datasets were the most important in different tasks. We suggest that multiple different mapping approaches should be tested and several remote sensing datasets used when maps of vegetation are produced. Saarimaa et al. 2019), water storage and hydrology (Waddington et al. 2015) and carbon exchange (Aurela et al. 1998, 2009Loisel et al. 2017). Within peatlands, many biogeochemical cycles, such as flows of carbon, water and nutrients, are linked to vegetation (Loisel et al. 2017;Lees et al. 2018;McPartland et al. 2019). For

show abstract

Assessing Different Feature Sets’ Effects on Land Cover Classification in Complex Surface-Mined Landscapes by ZiYuan-3 Satellite Imagery

Cited by 44 publications

References 38 publications

Multimodal and Multi-Model Deep Fusion for Fine Classification of Regional Complex Landscape Areas Using ZiYuan-3 Imagery

Multimodal and Multi-Model Deep Fusion for Fine Classification of Regional Complex Landscape Areas Using ZiYuan-3 Imagery

A Texture Classification Approach Based on the Integrated Optimization for Parameters and Features of Gabor Filter via Hybrid Ant Lion Optimizer

Detecting northern peatland vegetation patterns at ultra‐high spatial resolution

Contact Info

Product

Resources

About