Research on Body Mass Estimation Method of Koi Broodstock Base on Feeding State Image Recognition Technology

Abstract

Koi is graceful and colorful. In the process of breeding, monitoring the growth state and predicting the body mass of the broodstock are both important. The broodstock will surface on the water while being feed. At this time, the digital camera captures the feeding state image rapidly. The head and mouth of the broodstock show the semi-spindle shape and bright color significantly which have very high contrast with the water background in the image. Based on the method of OpenCV image visual, feature triangle was defined based on the extracted colorful semi-spindle shape automatically. The projection area of the triangle in the image was computed through the vector graphics calculation by computer. A great deal of measurement data was taken analysis on the correlation between the shape parameters and the body mass of the broodstock by MATLAB. A correlation analysis model was established between the projection area of the feature triangle and the body mass. Through this method, the body mass of the broodstock in the aquaculture pond was estimated automatically. The estimating error of the body mass was lower than 9 percent. This method will reduce the workload of collecting the growth state data during the breeding process of the broodstock greatly. And the method can also improve the accuracy of the body mass estimating effectively, which has important production and application value.