scholarly journals A Generalized Zero-Shot Learning Framework for PolSAR Land Cover Classification

2018 ◽  
Vol 10 (8) ◽  
pp. 1307 ◽  
Author(s):  
Rong Gui ◽  
Xin Xu ◽  
Lei Wang ◽  
Rui Yang ◽  
Fangling Pu
Keyword(s):  
Land Cover ◽  
Land Cover Classification ◽  
Land Cover Class ◽  
Classification Framework ◽  
Training Stage ◽  
Training Samples ◽  
Semantic Relevance ◽  
Cover Class ◽  
Semantic Attributes ◽  
Generalized Zero

Most supervised classification methods for polarimetric synthetic aperture radar (PolSAR) data rely on abundant labeled samples, and cannot tackle the problem that categorizes or infers unseen land cover classes without training samples. Aiming to categorize instances from both seen and unseen classes simultaneously, a generalized zero-shot learning (GZSL)-based PolSAR land cover classification framework is proposed. The semantic attributes are first collected to describe characteristics of typical land cover types in PolSAR images, and semantic relevance between attributes is established to relate unseen and seen classes. Via latent embedding, the projection between mid-level polarimetric features and semantic attributes for each land cover class can be obtained during the training stage. The GZSL model for PolSAR data is constructed by mid-level polarimetric features, the projection relationship, and the semantic relevance. Finally, the labels of the test instances can be predicted, even for some unseen classes. Experiments on three real RadarSAT-2 PolSAR datasets show that the proposed framework can classify both seen and unseen land cover classes with limited kinds of training classes, which reduces the requirement for labeled samples. The classification accuracy of the unseen land cover class reaches about 73% if semantic relevance exists during the training stage.

scholarly journals Pemanfaatan Citra Landsat Untuk Klasifikasi Tutupan Lahan Lanskap Perkotaan Kota Palu

2017 ◽  
pp. 58
Author(s):  
ANDI CHAIRUL ACHSAN
Keyword(s):  
Land Cover ◽  
Image Classification ◽  
Urban Landscape ◽  
Satellite Image ◽  
Land Cover Classification ◽  
Landsat Image ◽  
Land Cover Class ◽  
Vector Data ◽  
Land Cover Map ◽  
Cover Class

ABSTRACTThe Use of Landsat Image For Land Cover Classification On Urban Landscape Of Palu CityOne of effort that can be done for knowing urban development is with seeing land cover information. By land cover map can be kown the information of kind of land cover, area of each land cover and thedistribution of the use of space on the region. Satellite image is one of media that can be used for obtaining the information of land cover. This research purpose was for analysing the use of landsat image for determining the land cover classification on urban landscape using geographic information system. Classification proces of satellite image was started by merging band and cropping image, geometric correction and classification using image classification and the data from classification was converted from raster data to the vector data. From conversion result was obtained information of land cover class for every kind of land cover. Results of land cover class consist of building with area 3110, 79 ha, forest with area 1689,34 ha, farm with area 386,67 ha, and water body 830,35 ha.Keywords : landsat image, classification, land cover

scholarly journals The Aerosol Index and Land Cover Class Based Atmospheric Correction Aerosol Optical Depth Time Series 1982–2014 for the SMAC Algorithm

2017 ◽  
Vol 9 (11) ◽  
pp. 1095 ◽  
Author(s):  
Emmihenna Jääskeläinen ◽  
Terhikki Manninen ◽  
Johanna Tamminen ◽  
Marko Laine
Keyword(s):  
Time Series ◽  
Land Cover ◽  
Aerosol Optical Depth ◽  
Optical Depth ◽  
Atmospheric Correction ◽  
Land Cover Class ◽  
Cover Class ◽  
Aerosol Index

scholarly journals LAND COVER CLASSIFICATION ALOS AVNIR DATA USING IKONOS AS REFERENCE

2012 ◽  
Vol 9 (1) ◽  
Author(s):  
Bambang Trisakti ◽  
Dini Oktaviana Ambarwati
Keyword(s):  
Land Cover ◽  
Near Infrared ◽  
Good Accuracy ◽  
Confusion Matrix ◽  
Principal Component ◽  
Land Cover Classification ◽  
Accuracy Evaluation ◽  
Post Processing ◽  
Classification Result ◽  
Training Samples

Abstract.  Advanced Land Observation Satellite (ALOS) is a Japanese satellite equipped with 3  sensors  i.e.,  PRISM,  AVNIR,  and  PALSAR.  The  Advanced  Visible  and  Near  Infrared Radiometer (AVNIR) provides multi spectral sensors ranging from Visible to Near Infrared to observe  land  and  coastal  zones.  It  has  10  meter  spatial  resolution,  which  can  be  used  to map  land  cover  with  a  scale  of 1:25000.  The  purpose  of  this  research  was  to  determineclassification  for  land  cover  mapping  using  ALOS  AVNIR  data.  Training  samples  were collected  for  11  land  cover  classes  from  Bromo  volcano  by  visually  referring  to  very  high resolution  data  of  IKONOS  panchromatic  data.  The  training  samples  were  divided  into samples  for  classification  input  and  samples  for  accuracy  evaluation.  Principal  component analysis (PCA) was conducted for AVNIR data, and the generated PCA bands were classified using Maximum Likehood  Enhanced Neighbor method. The classification result was filtered and  re-classed  into  8  classes.  Misclassifications  were  evaluated  and  corrected  in  the  post processing  stage.  The  accuracy  of  classifications  results,  before  and  after  post  processing, were  evaluated  using  confusion  matrix  method.  The  result  showed  that  Maximum Likelihood  Enhanced  Neighbor  classifier  with  post  processing  can  produce  land  cover classification  result  of  AVNIR  data  with  good  accuracy  (total  accuracy  94%  and  kappa statistic 0.92).  ALOS AVNIR has been proven as a potential satellite data to map land cover in the study area with good accuracy.

scholarly journals Urban Land-Cover Classification Using Side-View Information from Oblique Images

2020 ◽  
Vol 12 (3) ◽  
pp. 390
Author(s):  
Changlin Xiao ◽  
Rongjun Qin ◽  
Xiao Ling
Keyword(s):  
Feature Extraction ◽  
Land Cover ◽  
Remote Sensing Data ◽  
Land Cover Classification ◽  
Side View ◽  
High Resolution Data ◽  
Training Samples ◽  
Surface Models ◽  
Extraction Algorithm ◽  
Oblique Images

Land-cover classification on very high resolution data (decimetre-level) is a well-studied yet challenging problem in remote sensing data processing. Most of the existing works focus on using images with orthographic view or orthophotos with the associated digital surface models (DSMs). However, the use of the nowadays widely-available oblique images to support such a task is not sufficiently investigated. In the effort of identifying different land-cover classes, it is intuitive that information of side-views obtained from the oblique can be of great help, yet how this can be technically achieved is challenging due to the complex geometric association between the side and top views. We aim to address these challenges in this paper by proposing a framework with enhanced classification results, leveraging the use of orthophoto, digital surface models and oblique images. The proposed method contains a classic two-step of (1) feature extraction and (2) a classification approach, in which the key contribution is a feature extraction algorithm that performs simplified geometric association between top-view segments (from orthophoto) and side-view planes (from projected oblique images), and joint statistical feature extraction. Our experiment on five test sites showed that the side-view information could steadily improve the classification accuracy with both kinds of training samples (1.1% and 5.6% for evenly distributed and non-evenly distributed samples, separately). Additionally, by testing the classifier at a large and untrained site, adding side-view information showed a total of 26.2% accuracy improvement of the above-ground objects, which demonstrates the strong generalization ability of the side-view features.

scholarly journals Domain Adversarial Neural Networks for Large-Scale Land Cover Classification

2019 ◽  
Vol 11 (10) ◽  
pp. 1153 ◽  
Author(s):  
Mesay Belete Bejiga ◽  
Farid Melgani ◽  
Pietro Beraldini
Keyword(s):  
Neural Networks ◽  
Land Cover ◽  
Transfer Learning ◽  
Large Scale ◽  
Domain Adaptation ◽  
Land Cover Classification ◽  
Target Domain ◽  
Training Samples ◽  
Latent Space ◽  
Large Scale Land

Learning classification models require sufficiently labeled training samples, however, collecting labeled samples for every new problem is time-consuming and costly. An alternative approach is to transfer knowledge from one problem to another, which is called transfer learning. Domain adaptation (DA) is a type of transfer learning that aims to find a new latent space where the domain discrepancy between the source and the target domain is negligible. In this work, we propose an unsupervised DA technique called domain adversarial neural networks (DANNs), composed of a feature extractor, a class predictor, and domain classifier blocks, for large-scale land cover classification. Contrary to the traditional methods that perform representation and classifier learning in separate stages, DANNs combine them into a single stage, thereby learning a new representation of the input data that is both domain-invariant and discriminative. Once trained, the classifier of a DANN can be used to predict both source and target domain labels. Additionally, we also modify the domain classifier of a DANN to evaluate its suitability for multi-target domain adaptation problems. Experimental results obtained for both single and multiple target DA problems show that the proposed method provides a performance gain of up to 40%.

scholarly journals THE USE OF ONLINE WEB-BASED SOFTWARE I-TREE CANOPY AS AN ALTERNATIVE IN SUPPORTING COMMUNITY BASED ZERO DELTA Q PLANNING

2018 ◽  
Vol 2 (2) ◽  
pp. 120
Author(s):  
Akhmadi Puguh Raharjo
Keyword(s):  
Land Cover ◽  
Planning Process ◽  
River System ◽  
Tree Canopy ◽  
Land Cover Class ◽  
Community Based ◽  
Web Based ◽  
The Difference ◽  
Cover Class ◽  
Online Web

Zero Delta Q is a policy to ensure that any additional surface runoff due to development does not further burden the drainage or river system. In case of Zero Delta Q application planning at the community level, a software is needed that can help classify and quantify the existing land cover class in area where the community is located. The purpose of this study is to look at the time needed and reliability of the i-Tree Canopy web-based software online in classifying and quantifying land cover classes on one of the sub-catchments in the Pesanggrahan River Basin. The land cover class is divided into six: trees, grasses / undergrowth plants, open area, water bodies, pavement / road and roof of the building. For comparison, an RBI map is used from the same area to see the extent of each class of land cover. Observation of each point requires an average time of 5.2 ± 1.0 seconds. The difference between direct sub-basin measurements using i-Tree Canopy and detailed analysis results from the RBI map is within the range of 0.41% or 0.36 Ha for each individual class of land cover. For a relatively small study area (under 100 ha) and when supported with reliable internet access, this web-based online software is sufficiently reliable in assisting the application planning process to support Zero Delta Q policy.

Impact of Training Set Size on Object-Based Land Cover Classification

2014 ◽  
Vol 5 (3) ◽  
pp. 49-67 ◽  
Author(s):  
Gerhard Myburgh ◽  
Adriaan van Niekerk
Keyword(s):  
Land Cover ◽  
Classification Tree ◽  
Land Cover Classification ◽  
Training Sample ◽  
Support Vector ◽  
Training Set ◽  
Training Sample Size ◽  
Set Size ◽  
Object Based ◽  
Training Samples

Supervised classifiers are commonly employed in remote sensing to extract land cover information, but various factors affect their accuracy. The number of available training samples, in particular, is known to have a significant impact on classification accuracies. Obtaining a sufficient number of samples is, however, not always practical. The support vector machine (SVM) is a supervised classifier known to perform well with limited training samples and has been compared favourably to other classifiers for various problems in pixel-based land cover classification. Very little research on training-sample size and classifier performance has been done in a geographical object-based image analysis (GEOBIA) environment. This paper compares the performance of SVM, nearest neighbour (NN) and maximum likelihood (ML) classifiers in a GEOBIA environment, with a focus on the influence of training-set size. Training-set sizes ranging from 4-20 per land cover class were tested. Classification tree analysis (CTA) was used for feature selection. The results indicate that the performance of all the classifiers improved significantly as the size of the training set increased. The ML classifier performed poorly when few (<10 per class) training samples were used and the NN classifier performed poorly compared to SVM throughout the experiment. SVM was the superior classifier for all training-set sizes although ML achieved competitive results for sets of 12 or more training areas per class.

Sign in / Sign up

Export Citation Format

Share Document