Image classification is perhaps the most important part of digital image analysis in remote sensing. It is divided into two different methods, supervised classification and unsupervised classification. 

Supervised classification is a dominant method in land-use and land-cover mapping and land cover change monitoring. However, the performance may vary depending on the different function of labeled training data properties, such as the sample size and the statistically unbiased and discriminative capabilities of the features extracted from the data. 

A team of scientists from the Xinjiang Institute of Ecology and Geography led by Jilili Abuduwaili present novel techniques designed to address heterogeneous domain adaptation problems, i.e., situations in which the data to be processed and recognized are collected from different heterogeneous domains. 

The novel techniques are supervised multi-view canonical correlation analysis ensemble and its semi-supervised version. 

Scientists use the multi-view canonical correlation analysis scheme to extract multiple correlation subspaces that are useful for joint representations for data association across domains. This makes homogeneous domain adaption algorithms suitable for heterogeneous domain adaptation problems. 

Their study showed that both of the proposed procedures outperform the ones used in the comparison with respect to not only the classification accuracy but also the computational efficiency. 

The research was published in Remote Sensing entitled “Supervised and Semi-Supervised Multi-View Canonical Correlation Analysis Ensemble for Heterogeneous Domain Adaptation in Remote Sensing Image Classification”.