Valuing the African Genius
Conceived and designed by Kumatoo
Debunking myths around African talent
Profile reviewed: January 2, 2017
First publication: November 2012
Ishmael Msiza is a South-
The fingerprint classification system that they have developed, also called "a structural fingerprint classifier", can correctly classify a fingerprint with only partial information. It is a better system if compared to conventional techniques used for fingerprint classification. Indeed, fingerprint recognition systems have two essential transactions which are:
1) verification transaction,
2) identification transaction.
A fingerprint verification transaction is simpler because the system compares the fingerprint presented by one user against a template associated with a specific identity claimed by the user. However, things become more complex when a user does not claim any identity. The system must then establish the identity of that user by looking for a likely match against a huge number of fingerprint templates within a database. This is where the problem lies as the recognition system can, for a large template database, take a lot of time to find a likely match.
The solution is in shortening the search process by sorting fingerprints into various classes. The advantage is that a database search can be launched within a specified class, instead of searching the whole database. It is in this area that the CSIR has made its breakthrough by developing a classification technology. What the CSIR has achieved is now to classify fingerprints, even if some points (called delta points) are missing.
With this new technology, it is possible to capture all types of fingerprints and correctly classify them, using the classification software developed by Ishmael Msiza and the team.
In short, the new fingerprint system is revolutionary. It speeds up the recognition of fingerprints; it is accurate, efficient and cost effective. But it is not yet on the market. Why? According to Ishmael Msiza, "The system has not been commercialized yet because we are still evaluating different intellectual property (IP) protection models. As soon as the ideal model is identified, we can then start making headway in that direction."