Cloud Technologies and Neural Networks Are the Future of Remote Sensing Industry
The number of the satellites increases every year, as well as the quality of received imagery data causing exponential growth of the archives. According to UCS Satellite Database, as of January 1, 2021, there are over 900 active ERS satellites in orbit, with 144 of them launched in 2020 alone. Daily volume of satellite imagery data is enormous. Traditional approaches to data processing are sub-optimal, therefore, over recent years companies in the ERS field have been increasingly focusing on automatizing data-related processes (from data acquisition on out to presenting final products to clients). Cloud technologies are the ultimate solution for working with remote sensing data in today's reality since they ensure stable and fast access to data, as well as provide scalable self-adjusting tools for data processing.
SCANEX has extensive experience in developing hardware and software products for remote sensing field. We actively promote automated processes within our own projects; for instance, SCANEX cloud geo-information system is currently at the final stage of development. Such system will allow user to order, receive, process, create derivative products, as well as visualize and publish results in modern geo-service formats in a semiautomatic mode. MultiScan Company (part of SCANEX Group) heads the project with the support of the Foundation for Assistance to Small Innovative Enterprises in Science and Technology (Innovation Assistance Foundation), contract No. 566ГРНТИС5/49486.
Several other SCANEX projects aimed at creating automatically thematic products, providing prompt access to said products, as well as developing an intuitive interface. For example, one of the projects involved developing an automated system for data obtaining, processing and its analysis, as well as generating resulting layers with probable changes in forest cover according to data from Sentinel-2 and Landsat-8 satellites. The service has cloud computing and machine learning algorithms at its core, which allows detecting with high degree of probability logging areas and fire scenes. It will be also possible in the near future to detect areas with forest dieback and wind-thrown trees. Russian territory is vast, thus, this task is of national importance. You can read more on how the algorithm works here. Another system automatically detects ice bodies and areas with difficult ice-conditions along the Northern Sea Route using radar imagery data and additional ship information.
SCANEX congratulates Central Research Institute for Machine Building and thanks for the opportunity to participate in the conference and share our own accomplishments.