HIGH GROWTH RATE AND LOW FEED CONVERSION RATES USING NEW TYPE DEMAND FEEDING SYSTEM WITH IMAGE PROCESSING PROGRAM AND FISH BEHAVIOR

Abstract

Demand feeding system serves feeds to fish, when fish switch on feeders. Although demand feeding system has advantages, it still has problems, e.g. hierarchy problem of fish school and system learning period problem for fish. New type of demand feeding system was develop to solve these problems using fish behaviour and image processing system. At first, behaviour experiment was conducted using the image processing software Roborealm to obtain the optimum parameter for computer program. Through the behaviour experiment, two typical behaviour patterns were detected. When fish was hungry, fish group came to the water surface (H: parameter >63%), and when fish was not hungry, fish came to bottom (L: <45%) of fish tank. These two parameters were obtained and were put into the computer program in the workstation. HD Wi-Fi camera continuously recorded the real-time fish behaviour in fish tank, and when fish group came to above the “Hâ€, then the command was sent from workstation to microcomputer to send the order to feeding device to feed on. The results of feeding experiment showed this system could provide pellets to fish day and night time equally following fish behaviour. This feeding system could provide the pellets to fish based on fish requirements. This system showed higher growth rate and lower FCR than other feeding systems (timer feeder and demand feeder using an infrared light sensor).