See also the technology page !
The LTU image analysis platform can be adapted to suit the characteristics of your organization's specific visual assets. The platform has a comprehensive API along with several SDK’s that allow for the development of custom applications around the platform.
There are a number of product functionalities that LTU is offering as "R&D products". These products include outstanding functionalities and can be fine tuned and adapted to an organization's specific needs and requirements. In many cases, these products can be integrated as add-ons to existing LTU Image-Seeker and Image-Filter platforms.
If any of the products presented below are of interest to your organization, please feel free to contact us.
- Photo Album Management
- Video applications
- Media monitoring
- MMS Filtering
- Parental Control Software
- Antipiracy applications
- Other applications
Photo Album Management : automatic keyword suggestion
A fast and robust image-based software that can automate keywording, classification and semantic categorization of visual content.
- Automation of the image classification process
-
Standardization of keywording methods based on existing annotated image archive
- Reduction of costs and time related to fully-manual indexing
From image mining to video mining : criminal, terrorist, and other suspicious activities are increasingly the subject of videos. Videos are used by perpetrators (distribution via the Internet, peer-to-peer file exchanges, and the mobile telecommunications network). Videos are also used by law enforcement investigators in the pursuit of evidence. LTU Technologies has extended its expertise in image analysis to this rapidly emerging need for video analysis.
|
|
LTU video analysis relies on the same core technology as the LTU Image-Seeker. Comparing videos involves two main issues : - Extraction of visual data from videos and comparing the data with a database of reference database of known images and videos. - Equipping the analysis infrastructure with sufficient processing power to handle the increased processing and storage requirements of video data.
|
In all advertising sectors, image technology enables automatic comparisons of published and broadcast advertising (print, web, television) with original creative production materials. Media monitoring solutions profit from the speed and accuracy of automatic image comparison.
Are your digital assets being deployed efficiently ?
Does the image you are about to license already exist in your repository ?
Are you storing duplicate images on your SAN ?
Is identifying the best image a time consuming task ?
Are you leveraging your existing metadata to help identify new image assets ?
Are you protected against potential copyright violations ?
High-volume image databases
LTU’s image technology enables the management of high volumes of images, avoiding duplicates and providing visual search facilities for finding all images with related content.
Automatic image annotation
The software can propose keywords after analyzing image content. Existing meta-data is taken into account, and the automatically proposed keywords may be confirmed by a human operator. Thus indexing activities are speeded up, saving human resources and processing time, yielding a high return on investment.
The Trademark and copyright monitoring
With the help of a web crawler, image analysis technology enables the monitoring of images published on the Internet : it is possible to detect illicit use of trademarks and copyrights.
Spam by SMS is a reality in a certain number of countries. MMS spam is yet to come. LTU provides the best filtering technology for both SMSC and MMSC. Please see the porn filtering explanations pages.
Today, porn images are everywhere. Trying to filter images by using just text is rarely effective. For instance, an image entitled “teddybear.jpg” provides no information about the content of the image. Thus, to filter an image effectively, one must view the actual content. LTU is currently developing a new pornography filtering solution, to be used either as a standalone safeguard or as a complement to text-based filtering engines.
No prior watermarking procedures are necessary for the LTU technology to recognize protected images.
Watermarks are invasive. They have to be tagged onto images prior to distribution (it is too late for already-published images) . They have a questionable efficiency - some stock photography companies complain of a success rate as low as 10% (source: Stockart.com). In contrast, DNA’s are not invasive – No prior watermarking procedures are necessary for the LTU technology to recognize protected images, unless images are signicantly altered (visually). The LTU Image DNA calculations make it possible to find cloned images or even similar images, the latter a task being impossible with standard MD-5 or SHA-1 algorithms (any change to an original file would result in a change to the file’s digital fingerprint, as calculated by these standard algorithms)
Life sciences : the LTU solution is being tested to scan large volumes of microscopic images to identify growth characteristics. High-Throughput environments are also targeted. LTU has built several demonstrators showing how visual similarity defined on medical images can help recover essential physiological parameters for a given query image, by comparing the query to a set of pre-existing manually-annotated
Fight against graffiti and tags : graffiti and tags represent degradations that cost millions. Transportations, cleaning private companies, local collectivities and police forces are involved in this sociologic pleague. LTU image recognition can help investigation, by integration in the information systems (surface to be cleaned, geo).
