Abstract
The classification of image time series has potential significance in the field of land-cover analysis with the increasing number of remote sensing images. The key problem of the classification of image time series is how to transfer the already available knowledge on the source domain to the target domain. Nevertheless, most of the existing methods do not consider the impact of different sample costs on the classifier during transferring. In addition, it is very difficult to collect reliable labeled samples with changed or unchanged categories between the source domain and the target domain in the case of a large number of training samples. In order to alleviate the above problems, we propose a cost-sensitive self-paced learning (CSSPL) framework with adaptive regularization for the classification of image time series in this paper. Considering that the costs of different samples cannot be completely equal to the classifier, different cost values are assigned to each type of error first, then we minimize the total cost to give the change detection classifier a preference on the unchanged class, aiming to reduce wrong label propagation from the source to the target image. Besides, an adaptive mixture weight regularizer is designed to automatically assign sample weight based on the loss value of the data set, which enables more reliable sample weights to be selected for training. Experimental results show that the proposed algorithm provides a set of reliable samples for the training of classifier and achieves a promising improvement on classification accuracy.
Highlights
T HE classification of image time series in the remote sensing field has aroused great interest in recent years [1]–[4], which has been applied in detecting the land cover change, monitoring vegetation dynamics and crustal deformation, tracking the growth of crops, etc. [5]–[9]
It can be observed that convolutional neural network (CNN), LapSVM and transfer component analysis (TCA) fail to distinguish between water and soil
The classification maps achieved by CDCNN and cost-sensitive self-paced learning (CSSPL) are the closet to the reference image
Summary
T HE classification of image time series (images taken from the same scene at different times) in the remote sensing field has aroused great interest in recent years [1]–[4], which has been applied in detecting the land cover change, monitoring vegetation dynamics and crustal deformation, tracking the growth of crops, etc. [5]–[9]. Because of the fixed orbit, the satellite repeatedly passes by a certain location and takes massive amounts of images at different times. If it is only processed manually, it will result in a waste of lots of labeling time and human resources. For the classification of image time series problem, it aims to classify the newly acquired unlabeled images (target domain) by using the samples acquired on the same scene but at different times which is already labeled (source domain). A simple way is to use the classifier directly trained on the source domain to classify the target domain These kinds of approaches perform poorly in most cases [10]. Considering the different cost of false-negative (C1) and falsepositive (C−1), (1) can be rewritten as
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
