The objectives of this project are:

1. to develop an operational automatic identification system that will allow non-specialists to identify species of a target group of insects, and thus permit wide participation in processes designed to inventory and monitor biodiversity.

2. to develop the skills necessary in the partner country that will enable our collaborators to use the software to create automatic identification systems for other organisms.

3. to demonstrate to the broadest possible audience a novel British developed technique for overcoming the taxonomic impediment to the implementation of Article 7.

4. to provide a simple and cost effective method that will allow wide accessibility to diagnostic image-based specimens held in a UK collection.

---

Showing a specimen of an Ophionine parasitic wasp of the genus Enicospilus sp. at light.

The taxonomy of the group in question, Ophioninae, is reasonably sound and a traditional key is available for identifying the fauna throughout the region (Gauld 1988). However, species discrimination of ophionines is extensively based on minute differences in the shape of wing parts, differences that subsequently have proved to be very difficult for a non-specialist to appreciate. Consequently, in the various faunal surveys being undertaken by university students and others, great difficulty is experienced by people attempting to identify species using this key, and their success rates are very low. The problem that confronts us is not one of inadequate taxonomy (although there are some new species), it is one of user inexperience. It results from the fact that non-specialists are simply not able to perceive the very fine differences used for species segregation. This is a very common difficulty with taxonomic products, and is a barrier that is perhaps not widely appreciated unless taxonomists work with groups of users. Years of experience hone a taxonomists ability to perceive subtle differences in shape or curvature far beyond the level most untrained persons have. Clearly there is a need for a new type of product with which to do identification, one that does not require the user to gain a great deal of familiarity with the target organisms before they can confidently identify them. Modern computer technology, using machine vision techniques, offers a potential solution to this problem. A UK team has developed a pilot automatic identification system, capable of discriminating species based on non-apparent differences in wing morphology (Weeks et al., 1997a & b ).

This system would seem to be ideally suited for a group of insects with apparent, but subtle wing differences, and it would seem to hold great promise for overcoming the taxonomic impediment hindering the inventorying and monitoring of biodiversity. It is highly probable that the system will also be ideal for the screening and identification of other biological objects. For example, the system could be used to identify plants using their leaves. In addition, the generic and innovative pattern recogntion approaches adopted by the DAISY system will mean that it can be used in other areas where fuzzy pattern recognitions is needed, for example within medical physics (e.g. routine screening of cytological preparations), and within security systems (e.g. identification of individuals using retinal, iris and/or fingerprint analysis). Other possible applications are as diverse as aircraft contrail recognition and the location and subsequent tracking of tropical cyclones.