Fish freshness detection by digital image processing

Fish is a great food to consume. It is healthy, delicious, but has very limited shelf life. After that limited time, consuming the fish can damage people?s health. Fish also start involving safety aspects such as being free from harmful bacteria, parasites or chemicals.In order to decide whether a fish is fresh or not, there are some tips like checking it?s eyes, smell or gills. All of these methods require experience, detailed knowledge about fish and give results that are changing from person to person.In this thesis, a practical way of determining the freshness of fish is proposed. Photographs of fish are taken by mobile phone?s camera and no extra sensors are used. Images are processed using MATLAB and some informative features are extracted. With the use of those features, an artificial neural network decides the freshness of the fish.With the proposed technique, four different features are obtained to determine the freshness of the fish, three of which are related with the shape and the other is related to the colour. By using the features obtained from the shape, 90% of the test fish have been correctly recognized. With the addition of the colour feature all the test fish have been successfully classified. These results prove that the proposed technique can be easily and successfully used to detect the freshness of the fish.