|Author(s)||Mahe Kelig1, Parisi Vincenc2, Carbini Sebastien3, Soria J.A.2, Harbitz A4, de Pontual Helene3, Cotano U5, Songer S6, Millner R6, Lewy P7, Gudmundsson E.F8, Fablet Ronan9, Benzinou Abdesslam|
|Affiliation(s)||1 : Ifremer-150, quai Gambetta, BP 699, 62321 Boulogne sur mer, France
2 : Universitat Politècnica de Catalunya,Rambla Exposici6 59-69,08800 Vilanova i la Geltru, Spain
3 : Ifremer, Technopôle Brest-Iroise, BP 70,29280 Plouzané, France
4 : Institute of Marine Research,Postboks 6404,9294 Tromso, Norway
5 : AZTI,Herrera Kaia zig, 20110 Pasaia, Spain
6 : CEFAS, Pakefield Road, Pakefield, Lowestoft, Suffolk NR33 OHT, England
7 : DTU Aqua, Charlottenlund Slot, DK-2920, Denmark
8 : Marine Research Institute,Skulagata 4, P.O. Box 121 Reykjavik, Iceland
9 : Institut Telecom, Technopôle Brest-Iroise, 29238 Brest Cedex 3, France
10 : ENIB, Ecole Nationale d'Ingénieurs de Brest Laboratoire RESO (EA 3380) CS 73862 29238 BREST Cedex 3 - France
|Meeting||4th International Otolith Symposium, 24-28 August 2009 Monterey, California, USA|
|Abstract||Most of the fish stocks are assessed using age-based models, however age estimations using otoliths costs several million euros annually. In this context, Automatic Fish Ageing (EU Project) would provide means to standardize ageing among laboratories and build interpreted image databases ensuring the information conservation. Automation improves growth studies while reducing the cost of the acquisition of age data.
The whole processing chain from the acquisition of otolith data to the actual ageing issue using pattern recognition or statistical inference has been realised for three case studies : a) Cod (Gadus morhua ; Faeroes, North Sea, North East Arctic) b) Anchovy (Engraulis encrasicolus ; Bay of Biscay) c) Plaice (Pleuronectes platessa, Eastern Channel, Iceland).
A total of 6729 otoliths have been digitized and interpreted by the European readers.
An image processing tool for the extraction of geometric information in otolith images (algorithms for nucleus and reference growth axes detections, for the 2D segmentation of partial otolith rings and contrast-invariant representation of otolith ring) was developed while being relied on the reconstruction of individual otolith shape histories from otolith images. The results of the automatic estimation of individual age showed agreement percentages with the age estimated by the readers from 90.9% (Iceland Plaice) to 33.2% (North East Arctic Cod).
Conditional and mixture models were applied for automatic estimation of age structure starting from the features of fish (TL and W) and otolith (TL, W, Area, Major and Minor Axis Length, Perimeter).
Automatic Fish Ageing was integrated in a software TNPC 5.0 developed by the Noesis society company allowing an automated system for the acquisition of otolith images series and age estimation.
A cost/benefit analysis was carried out between these 3 automatic methods and the traditional ALK.
Recent advances in computer vision allows more reliable methods to extract information from otoliths and to interpret these features in order to estimate the age and growth of fish. But these methods should not be seen as being able to fully substitute experts excepted in trivial cases. They should rather been seen as tools to provide automatically extracted information that requires a subsequent control by experts both for age and growth estimations.
Mahe Kelig, Parisi Vincenc, Carbini Sebastien, Soria J.A., Harbitz A, de Pontual Helene, Cotano U, Songer S, Millner R, Lewy P, Gudmundsson E.F, Fablet Ronan, Benzinou Abdesslam (2009). Automatic Fish Ageing (AFISA). 4th International Otolith Symposium, 24-28 August 2009 Monterey, California, USA. https://archimer.ifremer.fr/doc/00011/12270/