Artificial Intelligence and Earth Observation: On-Board Pre-processing, Data Compression and Image Selection

Dear Colleagues,

Earth observation missions have progressively increased in recent years, providing easier access to space and the generation of incredible amounts of data; examples include the Copernicus missions, which produce 12 TB of data per day. The reasons for this success are manifold, but include advances in mini and cube satellite formations, the advent of digital multichannel technology, increase in computational power and decrease in the cost for onboard hardware.

However, there are two major impairments in the generation and cumulation of space mission data:

• The onboard storage capability and finite bandwidth of data downloads; counteracting such limits implies the use of costly solutions, such as wider onboard storage or more on-ground stations.
• The generation of unseen data; many products or acquired areas have likely never been used due to their redundancy or scarcity of interest, such as optical/NIR images that include a high percentage of clouds.

The second point is particularly important, since it paves the way for a multitude of algorithms aimed at recognizing, selecting, and saving only datasets that are of interest according to specific user requirements. These classification algorithms have recently experienced increased attention in the scientific community, thanks to the advent of artificial intelligence algorithms.

Previous Special Issues have focused on specific aspects of the AI processing of remote sensing data, such as for urban environments, or have remained quite general. This Special Issue aims to obtain a clear picture of the current state-of-the-art and possible trends in upcoming years of the application of onboard AI algorithms, with the purpose of compacting the data before downloading.

The proposed themes all involve onboard data compression, although pertain to different aims. Potential topics include, but are not limited to:

• Image selection for ad hoc data compression. The selection of specific images for onboard data compression. Based on the target (in SAR by reflectivity, polarization, incidence angle; in optical/NIR by geographical area, presence of clouds, etc.), a more efficient data representation can be obtained by searching for the most performance quantizer and the ad hoc tuning of inner quantization parameters. This may be relevant, as an example, for future SAR missions with digital multichannel antenna.
• Onboard preprocessing. Smart data preprocessing for efficient onboard data compression. The transformation of data to provide a correlation, for example, range compression for SAR data, or towards another sparse domain, could help AI to find optimal space tessellation and compact data representation.
• Onboard data compression for specific targets. AI algorithms and onboard processing could be exploited for the finding of novel and more compact data representations, especially for specific targets such as ship recognition in maritime environments in SAR image acquisition, which is also an interesting example of sparse signals.

Dr. Pietro Guccione
Dr. Luisa Verdoliva
Dr. Michele Martone
Prof. Dr. Xiaoxiang Zhu
Guest Editors

Manuscript Submission Information

Manuscripts should be submitted online at www.mdpi.com by registering and logging in to this website. Once you are registered, click here to go to the submission form. Manuscripts can be submitted until the deadline. All submissions that pass pre-check are peer-reviewed. Accepted papers will be published continuously in the journal (as soon as accepted) and will be listed together on the special issue website. Research articles, review articles as well as short communications are invited. For planned papers, a title and short abstract (about 100 words) can be sent to the Editorial Office for announcement on this website.

Submitted manuscripts should not have been published previously, nor be under consideration for publication elsewhere (except conference proceedings papers). All manuscripts are thoroughly refereed through a single-blind peer-review process. A guide for authors and other relevant information for submission of manuscripts is available on the Instructions for Authors page. Remote Sensing is an international peer-reviewed open access semimonthly journal published by MDPI.

Please visit the Instructions for Authors page before submitting a manuscript. The Article Processing Charge (APC) for publication in this open access journal is 2700 CHF (Swiss Francs). Submitted papers should be well formatted and use good English. Authors may use MDPI's English editing service prior to publication or during author revisions.

• onboard data compression
• artificial intelligence
• neural network
• synthetic aperture radar
• multi/hyperspectral remote sensing
• earth observation