Application of distributed SVM architectures in classifying forest data cover types

Trebar, Mira; Steele, Nigel

doi:10.1016/j.compag.2008.02.001

Cited by 30 publications

(14 citation statements)

References 6 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…On the other hand, the RBF kernel has less numerical computation difficulties [16] . Some applications have proved that the SVM, compared with the neural networks, detects the study targets more accurately with a lesser false alarm rate [5,17,18] . As shown in Fig.…”

Section: Svm Classificationmentioning

confidence: 99%

Monitoring dynamic change of land cover based on SVM and satellite images in Hanoi, Vietnam

Zhao

LIU

2010

Wuhan Univ. J. Nat. Sci.

View full text Add to dashboard Cite

An integral method, combining support vector machine (SVM) with remote-sensing analysis techniques, was explored to monitor Hanoi's dynamic change of land cover. The landsat thematic mapper (TM) image in 1993, the enhanced thematic mapper plus (ETM + ) image in 2000, and the image with the charge-coupled device camera (CCD) on the China-Brazil earth resources satellite (CBERS) in 2008 were used. Six land-cover types, including built-up areas, woodland, cropland, sand, water body and unused land, were identified. The detected results showed visually the rapid urban expansion as well as land-cover change of Hanoi from 1993 to 2008. There were 12 637.54 hm 2 cropland decreased between 1993 and 2000, and 8 227.6 hm 2 cropland decreased between 2000 and 2008. Compared with cropland, woodland firstly decreased and then increased, and the other types did not change significantly. The results indicate that CBERS dataset has the application potential in world resources researches.To a considerable degree, land-cover dynamic researches have been playing an important role for analyzing human activities [1] . As an advanced earth-observation means, remote sensing has a couple of advantages. In different scales, it can be used to establish land-cover datasets [2] . For land-cover change researches, changes are usually monitored by comparing multiple images of the same ground area acquired at different dates. As a result, dozens of pattern recognition methods were developed. In addition to commonly used supervised and unsupervised classification techniques, attention has been focused on machine-learning algorithms, such as artificial neural networks (ANNs), since they use all spectral channels and provide complete information about land-cover changes [3] . Recent researches have indicated that support vector machine (SVM)-based approaches have the considerable potential for the supervised classification of remotely sensed data. Comparative studies have shown that SVM-based change detection can be more accurate than popular contemporary techniques such as neural networks, decision trees as well as conventional probabilistic classifiers [4,5] .Since early 1990s, Hanoi's land cover has changed dramatically. A large area of agricultural land was converted to be used for other purposes. Consequently, it is quite necessary to monitor dynamic change of land cover for supporting the decision-makers to build an appropriate land-use plan. Remote sensing was firstly introduced into Vietnam in the 1980s; it has been applied in many fields since then. In these applications, due to the multi-temporal

show abstract

Section: Svm Classificationmentioning

confidence: 99%

Monitoring dynamic change of land cover based on SVM and satellite images in Hanoi, Vietnam

Zhao

LIU

2010

Wuhan Univ. J. Nat. Sci.

View full text Add to dashboard Cite

show abstract

“…We determined C and ε first. When the best values of C and ε were set, the value of d 2 was selected according to the grid-search method, which was performed on C and ε using 10-fold cross-validation because the method could prevent the over fitting problem (Trebara & Steele, 2008). Taking the Gaussian kernel and the case 2 in 3.4.1 as kernel and input parameters of the SVR model, a detailed description of grid-search process is shown in Table 5, in which the optimal pair C ¼ 2 5 and ε ¼ 2 À3 was found with a cross-validation rate of 91.4%.…”

Section: Svr Parameters Optimizationmentioning

confidence: 99%

Prediction of TVB-N content in eggs based on electronic nose

Liu

2012

Food Control

View full text Add to dashboard Cite

“…In the particular context of hyperspectral imagery, some proposals are adjustments of conventional learning algorithms [2,16,28], whereas others use classifier ensembles [24,27] or feature selection techniques [5].…”

Section: Introductionmentioning

confidence: 99%

Classification of High Dimensional and Imbalanced Hyperspectral Imagery Data

García

Sánchez

Mollineda

2011

Pattern Recognition and Image Analysis

View full text Add to dashboard Cite

Abstract. The present paper addresses the problem of the classification of hyperspectral images with multiple imbalanced classes and very high dimensionality. Class imbalance is handled by resampling the data set, whereas PCA is applied to reduce the number of spectral bands. This is a preliminary study that pursues to investigate the benefits of using together these two techniques, and also to evaluate the application order that leads to the best classification performance. Experimental results demonstrate the significance of combining these preprocessing tools to improve the performance of hyperspectral imagery classification. Although it seems that the most effective order of application corresponds to first a resampling algorithm and then PCA, this is a question that still needs a much more thorough investigation.

show abstract

Application of distributed SVM architectures in classifying forest data cover types

Cited by 30 publications

References 6 publications

Monitoring dynamic change of land cover based on SVM and satellite images in Hanoi, Vietnam

Monitoring dynamic change of land cover based on SVM and satellite images in Hanoi, Vietnam

Prediction of TVB-N content in eggs based on electronic nose

Classification of High Dimensional and Imbalanced Hyperspectral Imagery Data

Contact Info

Product

Resources

About