.png){kind=icon}
Return
Edge Gradient-Based Active Learning for Hyperspectral Image Classification
.png){kind=icon}
PRE
AI
DOI：10.1109/LGRS.2019.2951800.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
In active learning (AL)-based remote sensing (RS) image classification tasks, the acquisition of labeled data depends not only on the informativeness and representativeness measured in feature space but also on the spatial distributions and relations in an image plane. However, very few studies have investigated the advantages of integrating spatial constraints into the AL paradigm. Hence, under the basic assumption instances that are difficult to classify are usually located around edges between different objects or land-cover types, edge gradient information was integrated into the conventional AL paradigm using popular uncertainty and diversity measurements. The experimental results with two real hyperspectral images confirmed the advantages of the proposed edge gradient-based AL (EGAL) approach from the aspects of fast convergence and computationally efficient operation.
Keywords:
Hyperspectral imaging
Training
Uncertainty
Measurement uncertainty
Support vector machines
Image edge detection
Active learning (AL)
edge gradient
image classification
informative sampling
support vector machine (SVM)
AI Summary
Key information extracted from the uploaded paper, including a brief overview, abstract, background, key highlights, visual analysis, and future outlook.
Journal
IF:
16.4
Papers:
1.0W
Citations:
5.1K
Organization
View more organizations
Cited Papers
.png){kind=icon}
ShareSaveShareSaveShareSave View more