PREDICTING SALMON BIOMASS REMOTELY USING A DIGITAL STEREOIMAGING TECHNIQUE

Effective fish farm management requires accurate information on fish b iomass in order to control feeding regimes, stocking densities and ult imately the optimum time to harvest the stock. Current methods for bio mass estimation are deemed inaccurate and may also be stressful to the fish, Therefore, the ability to remotely predict fish weight and biom ass was tested using a non-invasive, digital stereo-camera system. The camera system was arranged in a vertical set-up, which grabbed images of fish viewed from the side. Specific combinations of fin-fin, body depth and length dimensions, visible from side views of fish, were mea sured from stereo-images and these estimates were incorporated into a series of multifactor regression equations that were used to predict w eight. Biomass of Atlantic salmon (Salmo salar L.) was estimated to wi thin 0.4% of the real value and individual weight was determined with an error of -0.1 +/- 9.0%. Thus, the results from this study suggest t hat biomass can be predicted with a high degree of accuracy using a st ereo-camera system, and the method has the advantage of greatly reduce d stress levels in fish, compared with current biomass estimation tech niques.