Skip to main content

On board Processing for Compression and Clouds Classification in hyperspectral data

Periodic Reporting for period 1 - OP3C (On board Processing for Compression and Clouds Classification in hyperspectral data)

Reporting period: 2015-01-01 to 2015-06-30

Satellite hyperspectral sensors are instruments that deliver very high volume images. The several hundred spectral bands available, provide a very detailed prospective of the observed objects. These sensors have numerous applications ranging from environmental mapping to object identification (defense/security applications). The industry related problem is the technical constraints of the satellites. They have limited bandwidth transmission rates and limited storage and processing capabilities on board. These constraints result to either reducing the acquisition frequency or reducing the spectral resolution of the acquired images. Also since there is no knowledge about the cloud coverage, many images will be acquired and contain high percentage of cloud coverage and be of no use. As expected these have a negative impact on the potential and operational applications derived from these data. This is translated to economic loses both to the data providers/distributers and to the added-valued chain of products, since fewer or non-useful images are acquired and the end-user target group shortens.
The business opportunity lies on the above-mentioned technical constraints. By providing a solution to these problems, the industry will gain multiple benefits (acquisition of more images, down-streaming of the most appropriate images, engaging new application domains or extending existing ones) and grow its turnover.
The objective of the overall innovation project is to industrialize either a software or a FPGA version of the developed UNPCA compression and cloud classification algorithm for on board use on satellites containing HSI sensors (hyperspectral, ultraspectral, infrared sounding). It will compress and map clouds in the images acquired from the HSI satellites simultaneously. The result will be the compressed image, the cloud coverage mask, and a compression structure for direct processing without decompression. Images (or their proper cropped sub-sets) can be automatically selected according to pre-set criteria, or the operator can decide if they suite to his needs, in terms of cloud cover, for down-streaming case by case.
By providing this kind of innovative solution, the industry will gain multiple benefits (acquisition of more images, down-streaming of the most appropriate images, engaging new application domains or extending existing ones, lowering data transmission costs) and grow its turnover.
During the SME Instrument Phase 1 project, the work has been focused in the following categories: i.) Description and specifications of the methodology and the derived products, ii.) Technical analysis; Requirements/Constraints for FPGA to satellite/sensor integration, iii.) Technical analysis; Steps and processes to reach TRL 8, iv.) Market analysis report, v.) Cost-Benefit analysis of the proposed solution/product, vi.) Interaction and networking with the users/customers, vii.) Business plan.
The first part of the work contained the description of the developed compression/decompression and the cloud classification algorithm. The second part, i.e. the Technical analysis; Requirements/Constraints for FPGA to satellite/sensor integration, included the current practices used in HSI satellites. The requirements and the constraints that an FPGA must have in order to be integrated in a satellite are also described. Based on this analysis the product is defined. The following work included the steps and processes needed to implement for the defined product to reach the Technological Readiness Level 8.
The market analysis included the description of the market and its main aspects. Numerical analysis is performed, i.e. size of market, growth rate, regional/international opportunities and initiatives, and future estimation of the economic side of the market. Follow this, a cost analysis is performed, i.e. human resources, capital, infrastructure, are evaluated in order to determine the total resources needed to fully develop the product in TRL 8. Also the benefits of such a success are evaluated both for the company and the customers.
Networking activities were performed in order to communicate with the potential customers and present them our vision, our product stauts and provide us feedback. The main objective of this work was to enhance the impact of the future product and used as an extra input for the “Business plan”. Finally the business plan was complied. It included all necessary steps: commercialization strategy, operational planning, SWOT analysis, product description, timeline, and resources needed to accomplish the innovation’s project objective.
The novelty of the proposed business project is the creation of a new series of products (SW and HW) that aim to be integrated to satellites containing hyperspectral instrumentation enabling them to down-stream more and useful data. The latter has an increasing value either as raw data or as processed data (added valued products).
The key market application is the creation of such product. This field, i.e. compression and on board processing of hyperspectral data, is at its very early beginnings. Our innovation project stands out for the following reasons: i.) our solution is optimized for the hyperspectral data only, ii.) it uses remote sensing dedicated techniques to provide simultaneous compression and cloud detection, iii.) the compressed images can be directly processed without decompressing (compressed sensing), iv.) the data are de-noised and very highly compressed. The added value of our solution for the customers is first the fact that they can transmit more images requiring significantly less bandwidth and time (the satellite down-streaming is a high cost service) and second that they can decide if the image should be downloaded from the satellite (based on the cloud cover).
The proposed series of products can have advancements over time. For example further R&D could be required to further optimize the algorithm/FPGA hardware to adapt more efficiently to the specific characteristics of the integrated sensor or to cover evolving user needs. The main point in developing or investing in such a project is the fact that the satellite operators can reduce their down-streaming cost significantly (approximately a magnitude of 10). Also the infrastructures needed for archiving the data can be reduced by a factor of 10. Finally the ability to process the compressed data directly enables the data-mining procedures in very large archives.
The main economic benefits for the users are based on solving of the above identified needs. The down-streaming cost of data from satellite is defined practically with three terms: volume (kb), time (sec), rate (kb/s). The first economic benefit for the users is the lower down-streaming volume (more than 20 times lower volume) of data (compressed data and only cloudless data). The time that is needed to down-stream one compressed image versus the same image but classically encoded is 12-22 times faster. Also since the volume of the transmitted data is far less than the traditional encoding, down-streaming can be done with lower rate. Beside these direct economic benefits, there are also several indirect benefits that can be translated to economic. Due to the significantly lower down-streaming data volume, the energy consumption is far less. This has a double effect, first the hardware needed to support the telecommunications can be constrained, resulting to a more economic construction of the satellite. Second, the image acquisition tasking will be more flexible, since more energy will available for maneuvering the satellite. This results to more custom image acquisitions that are directly requested from the clients of our users, increasing their economic benefits of our users.
OP3C Cover Image
OP3C Press Release Image