Research Of Agricultural Pests Remote Automatic Monitoring System Based On Image Recognition

Agricultural pests is the biggest enemy of the agricultural production, especially with the rice pests, rice is one of the main food of our country, production of rice influence the national stability and people’s lives. The past for a period of time, due to the excessive use of pesticides, not only serious impact on the ecological environment also give own security brought serious hidden trouble. And because all Organism have their own evolution, a lot of insects resist to pesticides continuously strengthened, now needs to use the drug may be the past2times, three times or more, that caused the severe nausea cycle. Many studies show that scientific and effective information of species pests, distribution density can not only the correct use of pesticides, reduce the dosage of pesticides, even through the physical methods to prevent the outbreak of pests in advance.In order to timely and effective access to agricultural pests information, this paper on the basis of traditional trap devices, put forward by use of the image processing technology for agricultural pests remote automatic monitoring, this paper mainly studies the following four aspects:(1) The completion of the design of the whole system, according to the actual situation of the worm in the large number of body, to design shaking device and sampling machine device, to control of the worm quantity and quality, design with transportation to the worm, and transportation body brush up and down the way to solve the problem of stranded worm body. But also increase the small scope climate acquisition system for agricultural experts, provides climate data analysis pest problems (2) In view of the actual situation of the photography in the images have many targets,, the effective implementation of the pretreatment scheme. Through to the original pests of image enhancement, binary image, the target partition preprocessing steps and so on. Got to realize binary image in favour of the recognition and classification of single target.(3) Use morphological principle, in based on binary image, through the analysis of20original characteristics, choose the seven to analysis the characteristics of rice pests.(4) With the improvement of the conjugate gradient method to extract pests identification with characteristics, the learning efficiency than the traditional BP neural network has improved greatly, and the accuracy of test sample identification for85%.