Machine Learning Comparison between WorldView-2 and QuickBird-2-Simulated Imagery Regarding Object-Based Urban Land Cover Classification

Novack, Tessio; Esch, Thomas; Kux, Hermann Johann Heinrich; Stilla, Uwe

doi:10.3390/rs3102263

Cited by 102 publications

(59 citation statements)

References 28 publications

(32 reference statements)

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…In the process of hierarchical classification, RF uses a subset of features to build each individual hierarchy, making a more accurate classification decision based on effective features and reducing the error associated with the structural risk of the entire feature vector. Novack et al [54] showed that RF classifier can evaluate each attribute internally; thus, it is less sensitive to the increase of variables (Tables 5 and 6). The object-based classifier can provide faster and better results and can be easily applied to classify forest types [24,39,40,55].…”

Section: Discussionmentioning

confidence: 99%

Forest Type Identification with Random Forest Using Sentinel-1A, Sentinel-2A, Multi-Temporal Landsat-8 and DEM Data

et al. 2018

View full text Add to dashboard Cite

Abstract:Carbon sink estimation and ecological assessment of forests require accurate forest type mapping. The traditional survey method is time consuming and labor intensive, and the remote sensing method with high-resolution, multi-spectral commercial satellite images has high cost and low availability. In this study, we explore and evaluate the potential of freely-available multi-source imagery to identify forest types with an object-based random forest algorithm. These datasets included Sentinel-2A (S2), Sentinel-1A (S1) in dual polarization, one-arc-second Shuttle Radar Topographic Mission Digital Elevation (DEM) and multi-temporal Landsat-8 images (L8). We tested seven different sets of explanatory variables for classifying eight forest types in Wuhan, China. The results indicate that single-sensor (S2) or single-day data (L8) cannot obtain satisfactory results; the overall accuracy was 54.31% and 50.00%, respectively. Compared with the classification using only Sentinel-2 data, the overall accuracy increased by approximately 15.23% and 22.51%, respectively, by adding DEM and multi-temporal Landsat-8 imagery. The highest accuracy (82.78%) was achieved with fused imagery, the terrain and multi-temporal data contributing the most to forest type identification. These encouraging results demonstrate that freely-accessible multi-source remotely-sensed data have tremendous potential in forest type identification, which can effectively support monitoring and management of forest ecological resources at regional or global scales.

show abstract

Section: Discussionmentioning

confidence: 99%

Forest Type Identification with Random Forest Using Sentinel-1A, Sentinel-2A, Multi-Temporal Landsat-8 and DEM Data

et al. 2018

View full text Add to dashboard Cite

show abstract

“…[70][71][72][73], and (II) a full primal sketch, equivalent to perceptual grouping [33,52,80], where level 2 boundaries (e.g., texture boundaries) are detected between groups of tokens ( [13]; pp. 53,[91][92][93][94][95]. Unfortunately, Marr provided neither raw nor full primal sketch implementation details.…”

Section: About the Former A Rs-ius Cannot Detect Rs Image-objects Wimentioning

confidence: 99%

“…Unfortunately, image segmentation parameters are always site-specific and must be user-defined based on heuristics and a trial-and-error approach. For example, in the case of the popular Baatz et al segmentation algorithm adopted by the pre-attentive vision first stage of the Definiens GEOBIA/GEOOIA commercial software products [10], statistical methods have been developed to automatically optimize the parameters based on a site-specific training set of reference image-objects [9,85,93].…”

Section: Geobia/geooia Weaknesses (Due To Internal Drivers)mentioning

confidence: 99%

Operational Automatic Remote Sensing Image Understanding Systems: Beyond Geographic Object-Based and Object-Oriented Image Analysis (GEOBIA/GEOOIA). Part 1: Introduction

2012

View full text Add to dashboard Cite

According to existing literature and despite their commercial success, state-of-the-art two-stage non-iterative geographic object-based image analysis (GEOBIA) systems and three-stage iterative geographic object-oriented image analysis (GEOOIA) systems, where GEOOIA ⊃ GEOBIA, remain affected by a lack of productivity, general consensus and research. To outperform the degree of automation, accuracy, efficiency, robustness, scalability and timeliness of existing GEOBIA/GEOOIA systems in compliance with the Quality Assurance Framework for Earth Observation (QA4EO) guidelines, this methodological work is split into two parts. The first part of this work provides a multi-disciplinary Strengths, Weaknesses, Opportunities and Threats (SWOT) analysis of the GEOBIA/GEOOIA approaches that augments similar analyses proposed in recent years. In line with constraints stemming from human vision, this SWOT analysis promotes a shift of learning paradigm in the pre-attentive vision first stage of a remote sensing (RS) image understanding system (RS-IUS), from sub-symbolic statistical model-based (inductive) image segmentation to symbolic physical model-based (deductive) image preliminary classification. Hence, a symbolic deductive pre-attentive vision first stage accomplishes image sub-symbolic segmentation and image symbolic pre-classification simultaneously. In the second part of this work a novel hybrid (combined deductive and inductive) RS-IUS architecture featuring a symbolic deductive pre-attentive vision first stage is proposed and discussed in terms of: (a) computational theory (system design); (b) information/knowledge representation; (c) algorithm design; and OPEN ACCESSRemote Sens. 2012, 4 2695 (d) implementation. As proof-of-concept of symbolic physical model-based pre-attentive vision first stage, the spectral knowledge-based, operational, near real-time Satellite Image Automatic Mapper™ (SIAM™) is selected from existing literature. To the best of these authors' knowledge, this is the first time a symbolic syntactic inference system, like SIAM™, is made available to the RS community for operational use in a RS-IUS pre-attentive vision first stage, to accomplish multi-scale image segmentation and multi-granularity image pre-classification simultaneously, automatically and in near real-time.

show abstract

“…As the proposed pre-processing comes in addition and not as a replacement for the suggested procedures in Blaschke et al [12], the aggregation remains valid while integrating the important feature attribute of single pixel objects at the lowest GEOBIA level and using the properties at higher aggregation levels. Urban mapping and single building detection is a common topic in GEOBIA [15], and the use of convolution and edge detection is regarded amongst the more important features in building detection [16]. The equation presented extends the already available list of features where the ability to apply the analysis to very large neighborhoods is related to a trend in the availability of increased hardware capacities.…”

Section: Agricultural Application With Urban Mapping Extensionsmentioning

confidence: 99%

Spectral Difference in the Image Domain for Large Neighborhoods, a GEOBIA Pre-Processing Step for High Resolution Imagery

Kok

2012

Remote Sensing

View full text Add to dashboard Cite

Abstract:Contrast plays an important role in the visual interpretation of imagery. To mimic visual interpretation and using contrast in a Geographic Object Based Image Analysis (GEOBIA) environment, it is useful to consider an analysis for single pixel objects. This should be done before applying homogeneity criteria in the aggregation of pixels for the construction of meaningful image objects. The habit or "best practice" to start GEOBIA with pixel aggregation into homogeneous objects should come with the awareness that feature attributes for single pixels are at risk of becoming less accessible for further analysis. Single pixel contrast with image convolution on close neighborhoods is a standard technique, also applied in edge detection. This study elaborates on the analysis of close as well as much larger neighborhoods inside the GEOBIA domain. The applied calculations are limited to the first segmentation step for single pixel objects in order to produce additional feature attributes for objects of interest to be generated in further aggregation processes. The equation presented functions at a level that is considered an intermediary product in the sequential processing of imagery. The procedure requires intensive processor and memory capacity. The resulting feature attributes highlight not only contrasting pixels (edges) but also contrasting areas of local pixel groups. The suggested approach can be extended and becomes useful in classifying artificial areas at national scales using high resolution satellite mosaics.

show abstract

Machine Learning Comparison between WorldView-2 and QuickBird-2-Simulated Imagery Regarding Object-Based Urban Land Cover Classification

Cited by 102 publications

References 28 publications

Forest Type Identification with Random Forest Using Sentinel-1A, Sentinel-2A, Multi-Temporal Landsat-8 and DEM Data

Forest Type Identification with Random Forest Using Sentinel-1A, Sentinel-2A, Multi-Temporal Landsat-8 and DEM Data

Operational Automatic Remote Sensing Image Understanding Systems: Beyond Geographic Object-Based and Object-Oriented Image Analysis (GEOBIA/GEOOIA). Part 1: Introduction

Spectral Difference in the Image Domain for Large Neighborhoods, a GEOBIA Pre-Processing Step for High Resolution Imagery

Contact Info

Product

Resources

About