AFORO - Shape Analysis of Fish Otoliths



CMIMA

CSIC

UPC

VIC

Site best viewed with 1024 x 768 resolution

Publications and presentations



Publications:

> A. Lombarte, Ò. Chic, A. Manjabacas, R. Olivella, V. Parisi-Baradad, J. Piera and E. García-Ladona. Six years of the interactive AFORO (otolith shape analysis) database website (2003/2009) 2009. 4th International symposium fish otolith research and application. Monterrey (EE.UU).(PDF).

> A. Lombarte, M. Palmer, J. Matallanas, J. Gómez-Zurita and B. Morales-Nin. Ecomorphologic comparisons of otolith sagittae in Nototheniidae. 2009. 4th International symposium fish otolith research and application. Monterrey (EE.UU). (PDF).

> J.A. Soria, A. Lombarte, V. Parisi-Baradad. Otolith identification of Merluccius populations and sympatric species with local discriminant bases.2009. 4th International symposium fish otolith research and application. Monterrey (EE.UU). (PDF).

> E. Torrecilla, J. Piera, A. Lombarte and V. Parisi-Baradad. Automatic landmark selection of otolith shape contour using the Wavelet Transform Modulus Maxima.2009. 4th International symposium fish otolith research and application. Monterrey (EE.UU). (PDF).

> 2008. D. V. Lychakov, Y.T. Rebane, A. Lombarte, M. Demestre, L.A. Fuiman. Saccular otolith mass asymmetry in adult flatfishes. J. Fish. Biol. 72: 2579-2594. (PDF).

> 2007. A. Lombarte, A. Cruz. Otolith size trends in marine communities from different depth strata. J. Fish. Biol. 71: 53-76 (SCI: 1.393, citas: 5) (PDF).

> Lombarte, A., Ò. Chic, V. Parisi-Baradad, R. Olivella, J. Piera & E. García-Ladona. 2006. A web-based environment from shape analysis of fish otoliths. The AFORO database. Scientia Marina 70: 147-152 (PDF).

> Piera, J., V. Parisi-Baradad, E. García-Ladona, A. Lombarte, L. Recasens & J. Cabestany. 2005. Otolith shape feature extraction oriented to automatic classification with open distributed data. Marine and Freshwater Research, 56: 805-814 (PDF).

> Parisi-Baradad, V., A. Lombarte, E. García-Ladona, J. Cabestany, J. Piera & Ò. Chic. 2005. Otolith shape contour analysis using affine transformation invariant wavelet transforms and curvature scale space representation. Marine and Freshwater Research, 56: 795-804 (PDF).

> Cruz, A. & A. Lombarte. 2004.Otolith size and their relationship with colour pattern and sound production. Journal of Fish Biology, 65: 1512-1525 (PDF).

> V. Lychakov, Y.T. Rebane, A. Lombarte, L.A. Fuiman, A. Takabayashi. 2006. Fish otolth aymmetry: morphometry and modeling. Hearing Research 219: 1-11 (PDF).

Presentations:

3d International symposium fish otolith research and application:

> AFORO: An interactive shape analysis and classification system for fish otoliths (Abstract / Presentation).
> Otolith size trends in different depth marine communities (Abstract / Presentation).
> Otolith shape feature extraction oriented to artificial neural network classification (Abstract / Presentation).
> Otolith shape contour analysis using affine transformation invariant Wavelet Transforms and Curvature Scale Space representation (Abstract / Presentation).


Abstracts:

Oral Presentation: "AFORO: An interactive shape analysis and classification system for fish otoliths"
Authors: Chic, O. (1), Cruz, A. (1), Lombarte, A. (1), Olivella, R. (1), García-Ladona, E. (1), Parisi, V. (2), Graña, M. (4)

Abstract: Sagitta otolith shape variability was related with fish environment and their genetics. Otolith shape provides information of species, their ecobiological parameters and fish geographic origin.
Here we present the implementation of an interactive system to deal with shape analysis of fish otoliths and a classification system based on the mathematical properties of the one-dimensional curves describing the otolith contours. The system is connected to a database of complete morphometry information and otolith images of well identified samples. At present the database contains around 1000 high-resolution images corresponding to 200 species mainly from the Mediterranean and Antarctic Seas.
Queries are based directly on the numerical descriptors of otolith contours from their images. At present three main numerical descriptors have been implemented: FFT and wavelet spectrums and curvature scale space (SCC) representation of the otolith shape. As a first approach the classification strategy is based on a weighted algorithm of cluster analysis over the indexed numerical descriptors. The user may interactively change the search strategy according to the best descriptor. The otolith database, the shape analysis and the classification system are included into a web based environment, where a server links the database with the numerical routines to perform shape analysis, written using open source technology (Java, Postgress and Scilab) to ensure portability and development control.


Oral Presentation: "Otolith size trends in different depth marine communities"
Authors: Lombarte, A.(1), Cruz , A.(1)

Abstract: 660 sagitta otoliths from 132 species belonging to 7 demersal communities or subcommunities of different depth and bottom structure and one pelagic community from North Western Mediterranean were compared in order to study otolith relative size and morpho-functional trends. In every community was selected the most characteristic species. Sagitta otoliths were digitised from their medial side (sulcus acusticus side). The relationship between area of medial side of otolith sagitta and total length of the fish was calculated to determine otolith relative size. The otoliths were divided in three groups (small, medium and large) and was calculated the percentage of every otolith size group in each community.
The species from pelagic community were characterised by otolith of small and medium relative size. When compare demersal communities, the proportion of large and small otolith increase with depth (below 600-m depth). Instead, in the continental shelf and upper slope communities the relative medium sized otoliths were clearly most abundant than small and large otolith. A morpho-functional interpretation suggests an increase the relative importance of acoustic communication (related with large otolith size) in depth waters in order to compensate the limited visual field.


Oral Presentation: "Otolith shape feature extraction oriented to artificial neural network classification"
Authors: Piera, J.(3), Parisi, V.(2), Bermejo, S.(2), Cabestany, J.(2), García-Ladona, E.(1), Lombarte, A.(1)

Abstract: Otolith shape classification is a common procedure in otolith related studies (morphological, taxonomical, palaeontological and food web analysis).
This study reviews some of the critical pre-processing steps required for otolith shape classification in basis to Artificial Neural Networks (ANN): contouring, shape codification and shape feature extraction.
ANNs are powerful tools for automatic data classification. ANNs are non-linear models that develop weighted links between network processors (hidden neurons) and both data input values and target output classes. A common procedure for optimising ANNs classification is the application of data pre-processing, in order to reduce the dimension of vector inputs.
Sets of requirements for otolith image acquisitions are proposed in order to obtain a robust contouring procedure. Several codification methods (radial signature, complex contour and Freeman chain codes) are evaluated, pointing out the limitations (loss of information) and the benefits (invariance to affine transformations) associated to each method.
Three different types of descriptors are presented: morphological based, statistical based and spectral based. A comparative study of the shape descriptors is developed, focused on data reduction techniques for optimising ANNs classification.


Oral Presentation: "Otolith shape contour analysis using affine transformation invariant Wavelet Transforms and Curvature Scale Space representation"
Authors: Parisi, V.(2), Lombarte, A.(1), García-Ladona, E.(1), Cabestany, J.(2), Piera, J.(3), Chic, O.(1), Cruz, A.(1), Olivella, R.(1)

Abstract: Two decades ago different methods and techniques related to image analysis began to be used in fish otolith studies (ageing, stock determination, species identification ...). In this paper we show the application of two recent signal processing techniques (wavelet transform and curvature scale space), which complement the information gathered using Fourier analysis methods. The motivation comes from the necessity to perform an otolith analysis capable of localizing its contour singularities, since these are very important points in automated identification tasks, very close to landmark selection done by trained human operators.
The wavelet transform (WT) is computed by expanding a signal into a family of functions that represent the dilations and translations of a unique function known as a mother wavelet. As a difference to the Fourier transform, which is defined by an integral covering the whole signal, the wavelet transform is based on an analysing function located both in space and frequency, having the ability not only to measure the irregularities of the signal but also to establish its position.
The Curvature Scale Space (CSS) representation is another technique that maps the evolution of the inflection points of a shape contour when this is smoothed at increasing scales, providing a curve analysis invariant under scale, translation and rotation image changes.
The robustness of these techniques in otolith shape studies is illustrated through a classification system based on a distance measure. We test its performance to recognize otolith images stored in a database, under affine transformations, shear and in the presence of noise.
Additionally these techniques can be used for data compression purposes, which is very important for remote retrieval and broadcasting of large databases.


1.- Institut de Ciències del Mar (CSIC). Passeig Marítim 37-49, 08003 Barcelona, Catalonia, Spain.
2.- Dept. Eng. Electrònica, Universitat Politècnica de Catalunya (UPC), Barcelona, EU.
3.- Dept. Teoria del Senyal i Comunicacions, Universitat Politècnica de Catalunya (UPC), Barcelona, EU.
4.- Dept. de Ciencias de la Computación e Inteligencia Artificial. Univ. del País Vasco, San Sebastian (Spain).