A Greek innovative ICT company established in Athens, as part of the technology park of a large research center, has developed algorithms for software or hardware (SW/HW) co-design that can be integrated in earth observation cube satellites. The algorithms enhance data exploitation and enable critical applications such as the real-time object/change detection, real-time data screening and real-time monitoring for safety and security applications.
Apart from data acquisition and storage, the main data processing algorithms that have been heavily developed and deployed in satellite platforms are those that focusing on lossless compression of images. The aim is to address the challenge of the limitations posed by downlink bandwidth constraints and the limited viewing time of the ground stations. Several algorithms for multispectral and hyperspectral data (multispectral and hyperspectral imaging collect images of an object in a series of spectral windows with high spatial resolution) and data collected on red-green-blue images systems (RGB) are operationally deployed in all earth observation satellites.
The offered algorithms for software or hardware co-design include 3 modules. The first type is the onboard image quality evaluation. For the reduction of the storage and transmission requirements an initial onboard evaluation of the acquired data can be performed for assessing quantitatively the amount of clouds and shadows depicted in the acquired images. Above a certain threshold (e.g. more than 80% cloud coverage) the data can be discarded. The calculated quality layer can be exploited from operational geospatial services and for the automated screening of freshly acquired data.
The second type of processing is the onboard top of atmosphere at sensor radiance. This method aims to deliver market-ready products. For this purpose, the basic radiometric calibration and sensor corrections can be applied onboard to the acquired data, enabling significantly near real-time applications. Depending on the image size (spatial and spectral resolution) the image mosaicking can be also exploited to deliver relative new products or to further enhance onboard processing for offering products.
A third type of processing is the onboard object detection and change detection algorithms for critical geospatial applications. Currently, a major limitation of earth observation satellite data is the significant delay (from few to dozen days) from their acquisition to their availability on the cloud, minimizing their utilization capabilities and exploitation for several applications. Onboard, real-time data analytics can shorten the time between data acquisition and the necessary action and response, for instance, in case of disasters. To this end, two sub-types of onboard data processing algorithms can be considered towards addressing critical geospatial applications. The first one refers to object detection algorithms which involve vegetation detection and mapping, water detection and mapping (e.g. inland, coastal, marine), man-made detection and mapping (e.g. roads, buildings) and ship/vessel detection in marine environments. The second one is change detection algorithms which include change and no-change alerts, flood mapping and oil/debris marine detection.
The company seeks to find collaborators which need to integrate algorithms in earth observation cube satellites enhancing data exploitation, as well as enabling critical applications. The company is also interested in cooperating with partners that are already in this area in order to push the object detection software to production and/ or integrate it with larger systems. The types of collaboration sought are technical or research collaboration with various partners and provision of technical assistance from the Greek company.