The advances in processing performance and architectures have been impressive over the last ten years to the point that, in a number of configurations, the transfer of the data from Storage is the new bottleneck.
When the data to process is multi-source high resolution motion imagery, the volume of data to manipulate gets huge and cannot be reduced by the usual compression techniques as automatic content processing algorithms usually require the best possible imagery quality to achieve full performance.
If the data sources are dispersed and if only a small part of the content is expected to be of interest, there is an obvious mismatch in the allocation of the resources.
The objective of LINDO is to provide a new trade-off based on an architecture where storage is distributed and where, rather than moving content massively to central processing facilities, the relevant routines are sent to the remote sites to be run locally in each storage environment.
The results (metadata, indexes, content descriptions,…) are attached to the stored material for future use, but also made available to all the stake-holders. Then, only the minimum required content, extracted from its clip, can be transferred as required.
In this process, the distributed nature of the storage, of the indexing and of the processing is transparent to the end-user.
While the necessary content-based indexing tools have been the topic of many research projects, distributed generic architectures like LINDO have not been investigated in detail yet.
Established industries of Belgium, France and Spain independently noticed that the markets they serve were flooded by a myriad of video and multimedia sources and were in a growing demand of solutions.
Supported by research organizations of the same countries they concurred to initiate together with LINDO a future-proof approach: develop a generic architecture, standardize a minimum of four related interfaces and formats, and validate the result in real conditions with demonstrations in their most demanding market segments.
The booming videosurveillance applications, the broadcast archive access and the multimedia knowledge management will be addressed accordingl by the LINDO demonstrations; a number of other derived implementations are anticipated soon thereafter, an overall target market of several hundreds of millions of euros for the European Industry.