IEEE Geoscience and Remote Sensing Magazine - June 2016 - 32

CLASSIFICATION
Finally, the extracted feature is combined with the SVM
or another classifier to predict the scene label. However,
most methods for unsupervised feature learning produce
filters that operate either on intensity or color information.
Vladimir [113] proposed a quaternion PCA and k-means
combined approach for unsupervised feature learning that
makes joint encoding of the intensity and color information possible. In addition, Cheriyadat [58] introduced a
variant of sparse coding that combines local SIFT-based
feature descriptors to generate a new sparse representation,
producing an excellent classification accuracy. The sparse
AE-based method also produces excellent performance.
In [31], Zhang et al. proposed a saliency-guided sparse AE
method to learn a set of feature extractors that are robust
and efficient, proposing a saliency-guided sampling strategy to extract a representative set of patches from a VHR
image so that the salient parts of the image that contain
the representative information in the VHR image can be explored, which differs from the traditional random sampling
strategy. They also explored the new dropout technique in
the feature-learning procedure to reduce data overfitting
[114]. The extracted feature generated from the learned feature extractors can characterize a complex scene very well
and can produce an excellent classification accuracy.
sUPerVised hierarchicaL
FeatUre-LearninG-Based methods
Before 2006, it was believed that training deep supervised
neural networks was too difficult to perform (and indeed
did not work). The first breakthrough in training happened
in Geoff Hinton's lab with an unsupervised pretraining by
RBMs, as discussed in the previous subsection. However,
more recently, it was discovered that one could train deep
supervised networks by proper initialization, just large
enough for the gradients to flow well and the activations to
convey useful information. These good results with the pure
supervised training of deep networks seem to be especially
apparent when large quantities of labeled data are available.

Feature Extractor

In the early years after 2010, based on the latent Dirichlet
allocation (LDA) model [115], various supervised hierarchical
feature-learning methods have been proposed in the RS community [116]-[120]. LDA is a generative probabilistic graphical model for independent collections of discrete data and
is a three-level hierarchical model, in which the documents
inside a corpus are represented as random mixtures over a set
of latent variables called topics. Each topic is in turn characterized by a distribution over words. The LDA model captures
all of the important information contained in a corpus by
considering only the statistics of the words. The contextual
relationships are neglected due to the Bayesian assumption.
For this reason, LDA is categorized as a bag of words model.
Its main characteristic is based on the words' exchangeability.
The LDA-based supervised hierarchical feature-learning approaches have been shown to generate excellent hierarchical
feature representations for RS scene classification.
In fact, the LDA-based models are still not deep enough
compared to the other techniques in the DL family. More
recently, a few pure DL methods have been proposed for RS
image scene understanding based on CNNs [121]. Zhang et
al. proposed in detail a gradient-boosting random convolutional network framework for RS scene classification that
can effectively combine many deep neural networks [50].
Marmanis et al. considered a pretrained CNN by the ImageNet challenge and exploited it to extract an initial set of
representations for earth observation classification [122].
Hu et al. investigated how to transfer features from the existing successfully pretrained CNNs for RS scene classification [123]. Luus et al. suggested a multiscale input strategy
for multiview DL with the aid of convolutional layers to
shift the burden of feature determination from hand-engineering to a deep CNN [124]. These advanced supervised
DL methods all outperform the state-of-the-art methods
with the various RS scene classification data sets.
eXPeRIMents And AnALYsIs
In this section, we present some experimental results on
the DL algorithms for RS data scene understanding that we

K Channels
K Channels

Convolution
s

Input Image

Pooling

K * 16 Features

Pooling

Reshape

w
(n-w)/s+1
Feature Maps

Pooled Maps

Classifier Feature

fIgURe 6. The feature extraction using a w-#-w feature extractor and a stride of s. We first extract the w-#-w patches, each separated by s
pixels, then map them to the K-dimensional feature vectors to form a new image representation. These vectors are then pooled over
16 quadrants of the image to form a feature vector for classification.
june 2016

ieee Geoscience and remote sensing magazine

Table of Contents for the Digital Edition of IEEE Geoscience and Remote Sensing Magazine - June 2016