New paper: Automatic Selection of Relevant Attributes for Multi-Sensor Remote Sensing Analysis: A Case Study on Sea Ice Classification

In the new millennium, sea ice research has become a hot topic in the Earth observation disciplines, as it plays a vital role in the climate and polar ecosystem. Indeed, it affects several anthropogenic activities in the Arctic region, such as the oil and gas industry, fisheries, shipping, tourism, as well as the lifestyle and welfare of the indigenous population. Therefore, sea ice monitoring becomes a key interest in protecting the Arctic and ensuring safe and efficient activities and polar navigation.

PhD fellow Eduard Khachatrian is currently working on multimodal integrated remote sensing for Arctic sea ice monitoring. Nowadays, automatic sea ice charting is usually achieved by applying single-sensor imagery. However, it is fundamental to combine information from various remote sensing sensors with different characteristics in order to obtain more reliable sea ice characterization. Nevertheless, when analyzing images received from different sensors with both complementary and redundant characteristics, it is necessary to select an optimal set of image attributes that provides the relevant information content to enhance the efficiency and accuracy of the system.

In the following scientific article, the authors employed a fully automatic method based on Graph Laplacians for the selection of relevant attributes to separate different types of sea ice. Experiments show the advantage of applying multi-sensor data sets and demonstrate that the attributes selected by the proposed method result in high classification accuracies. The full-text version can be found here.