Cervical cancer screening
- Over 3,200 new cases each year in the UK
- Manual screening achieves only 56% identification
- Multi-purpose object recognition system:
– recognition = classification
– analysis of optical images, X-ray images, etc
- Demo software is available to detect cervical cancer (recognition of cervical cells)
- Can be applied to blood, thyroid, iris
- Uses a global knowledge database
- Can provide/share a platform for distance learning
- Algorithmic separation of cells from connected objects
- Fast cell nuclei identification
False documents identification
Our approach allows for the fast, simple and effective recognition of false documents (including passports). The key underlying idea is to compare the fractal structure of the blanc document with the over-script text. In case of any inconsistency, the scanner shows the problem. Several additional fractal features could be involved depending on application document. That is very easy to use even by those with minimal training required of police, customs, etc.
The scanning system has been successfully tested with British passports, and could be easily modified for other official documents.
The scanner is shown in the pictures below.
Rapid surface inspection
- Problems: High speed, missing defects, on line correction, grade specification.
- Makes use of multi-fractal analysis and fuzzy set theory:
– efficient structure analysis,
– fractal texture recognition,
– characteristic image properties are combined into a decision using fuzzy logic.
- Knowledge base provides for custom profiles to recognise (or ignore) specific defects.
- System could be customised for the similar inspection tasks.
Skin cancer screening
The ‘core technology’ is developed for a global integrated on-line screening service. The Expert System at the heart image recognition for each image of a suspect mole is analysed 42 different ways to produce a profile of both euclidian and fractal parameters.