Research And Development Of Automatic Bait Feeding Device Based On Embedded Machine Vision

With the progress of science and technology and national economy development, the aquaculture is developing from the extensive mode to the refinement. But for a long time, the feeders used traditional methods to feed, which took a long time, consumed much resource, and was low in efficiency, this situation limits the efficiency of aquaculture development. The traditional artificial feeding exists the high price of labor, feeding levels all depends on the subjective judgment consciousness, too high feeding levels may cause feeding waste and cause pollution to breeding environment, too low feeding levels may cause breed object growth uneven. So by using advanced embedded image processing technology this topic means develop aquaculture automatic feeding system. It is of great significance for the rational use of feed and improving the breeding benefit.At present, there is less basis study on combining the aquaculture project and embedded image analysis technology. The emphasis of this topic is mainly about statistical analysis of activity bait in aquaculture water and the dynamic analysis of breeding pond fish feeding. Main content can be summarized as the following aspects:(1) It is difficult to use traditional control methods to realize the precise control of feeding levels. So the theory of fuzzy control scheme is put forward for the characteristics of nonlinear, time-variant, multivariable coupling aquaculture feeding system. According to the principle of fuzzy control and structure, the fuzzy control rules is established to solve the problems. Finally, the experimental analysis of the fuzzy controller proves that fuzzy control can reduce the residual amount of bait and improve the breeding benefit.(2) This paper introduces the identification based on image processing method for floating bait, including the comparison of several kinds of segmentation algorithm in floating bait recognition effect, and analysis of the over-segmentation problem in watershed segmentation method. Realization of recognizing and counting the floating bait in image by using least squares support vector machine(LSSVM) and calculating the index of fish feeding activity is well.That relevant experimental platform has been established based on the key technology solved above and concrete experiment and analysis has been carried on. The results show that the floating bait automatic delivery system based on machine vision feeding is reasonable and improves the feeding efficiency and enhances the investment benefit.