Abstracts Category : Other

Computer vision system to detect salmon deformity

by Bjarni Bjarnason Jn

In the salmon farming industry, the salmon can be deformed, for example, in a circular shape or too thin. When the deformed salmon is gutted using automated machine, it can either cause the machine to stop or cause damage to the fish. To prevent this happening, human inspectors must decide whether the fish fits the machine. There are high salaries in Norway compared to other countries where other farming plants are, which provides a need for more automation. With more automation, Norway could be more competitive on the global market.The project presents a method to detect deformed salmon with a vision system. It includes measuring the geometric parameters of the fish, such as length, width, and angle of the back. By utilizing the parameters, a classifier was implemented to decide whether the fish is deformed. The software Halcon was used to make the image analysis program. laxeldis vinnslum er laxinn stundum afmyndaur annig a hann er hringlaga ea of grannur. egar afmyndaur lax er slgur sjlfvirkri vl getur hann stoppa vlina ea vlin skemmt laxinn. Til a koma veg fyrir etta skoar srstakt starfsflk fiskana ur en eir fara vlina. Laun Noregi eru hrri en rum lndum ar sem arar vinnslur eru og v er rf sjlfvirknivingu. Me v mti geta vinnslur Noregi ori samkeppnishfar alheimsmarkai. essu verkefni er lst afer til a greina afmyndaa laxa me myndgreiningar kerfi. Aferin felur sr a mla breytur, svo sem lengd, breidd og horn baki en einnig htt til a skera r um hvort fiskurinn s afmyndaur. Vi run aferinni var notast vi forriti Halcon til a ba til myndgreiningu sem notast er vi verkefninu. This project was done in collaboration with Valka ehf.