The Home Office automatic fingerprint pattern classification project

The purpose of the automatic fingerprint pattern classification project is to determine whether it is feasible to partition the main automatic fingerprint recognition database on the basis of gross fingerprint pattern type, without manual intervention. Such a procedure pre-supposes a satisfactory means of automatically recognising and categorising gross pattern type to a minimum level of accuracy. The work programme is designed to provide: an assessment of the performance of different neural network architectures, data representations and training methodologies in the automatic classification of fingerprint pattern types, and details of the most suitable fully trained neural network architecture for gross fingerprint pattern classification. A further objective is for the Home Office to be able to provide a demonstrable software implementation of a fully trained neural network algorithm