The identification of marine bivalve larvae using computational pattern recognition methods Jonathan Campbell 1, John Slater 2, John Gillespie 2, Ivan F. Bendezu 2 and Fionn Murtagh 3 1. Department of Computing, Letterkenny Institute of Technology, Port Road, Co. Donegal, Ireland. 2. Department of Science, Letterkenny Institute of Technology, Port Road, Co. Donegal, Ireland 3. School of Computer Science, Queen's University Belfast, Belfast, BT7 1NN
Marine Bivalve Larvae
Morphological Identification : Difficulties Microscopic ( m length)Microscopic ( m length) Shape (Egg-shaped)Shape (Egg-shaped) Colour (Brownish)Colour (Brownish) Lack of appendages (No legs, tentacles)Lack of appendages (No legs, tentacles)
Morphological Identification : Consequences Experienced personnelExperienced personnel Time-consuming for large sample nos.Time-consuming for large sample nos.
Morphological Identification : Solutions Automate - Pattern Recognition
Raw and Segmented Images
Image Processing Invariant Moments Shift invariant Shift invariant Scale invariant Scale invariant Rotation invariant Rotation invariant Data processing by statistical software package R Segmented image
Results – to date Scallop larvae shown in red, other larvae shown in black
Results – to date Scallop larvae shown in red, other larvae shown in blue Scallop
Conclusions Limted dataset investigated to dateLimted dataset investigated to date Automated identification of scallop larvae may be a possibilityAutomated identification of scallop larvae may be a possibility Need for further research with financial support involving a much larger data setNeed for further research with financial support involving a much larger data set