Research On The Method Of Lactation Piglet Group Target Detection And Behavior Feature Classification

Animal behavior is an important indicator of response to environmental comfort.In order to realize the automatic detection of captive lactation piglet survival environment temperature, this paper puts forward the use of digital image processing technology, features of lactation piglet behavior a new method of image classification and recognition, to distinguish between lactation piglet group under low temperature, suitable temperature, high temperature characteristics of behavior.The method can automatically identify more than accumulation of piglets, scattered, rest.After use of the target area and the Radon transform of the minimum bounding rectangle area ratio, and combining the between-cluster variance clustering method can quickly identify three characteristics of the behavior of the double threshold, thus effectively distinguish between the three characteristics of behavior.Through the image characteristics of 150 piglets lactation behavior test analysis, the result shows that the method characteristics of lactation piglet behavior recognition rate is respectively:cold,92%,86%,84% high temperature comfortable.The comprehensive recognition rate is higher than 87%.The research for the intelligent monitoring software in piggery temperature control provides data reference.