Research On Dynamic Image Feature Extraction In Precision Agriculture

The modern precision agriculture system is of great importance in resource conservation, environment improving, pollution reduction and improving benefit, which represent the development direction of agriculture in the future. In order to obtain agriculture field information efficiently, the interworking of multiple sensors, especially the acquisition and processing of visual information is needed in this process. With the rapid development of machine vision technology, high-definition image transfers the real-time of the field, but it also raises a new challenge to the extraction of dynamic characteristics. Corresponding to the specific project requirement in the precision agriculture, some solutions are proposed to extraction of dynamic image feature.To solve the problem of precision feeding for fish, in the nature field conditions, there have been many of the common problems of the complexity of aquatic environment, acuteness of target’s change, disturbance from the outside and so on. With the purpose of achieving dynamic objects’detectable quantification, definition of quantity of motion as well as the precision operating of the baiting system, according to the regular pattern of the target groups’ feeding activity, this study integrates and optimizes the detection method for the sake of the accurate calculation of target groups, which will attribute to achieving precision detection and feature extraction of dynamic targets. The mentioned method of optimizing target group’s point set used to relative feature detection not only improves the arithmetic efficiency, but it also guarantees a sustainable and stable operation of the system.To solve the problem of pest detection, the work is mainly about quantity detection, feature extraction and identification and classification. And he strategies of analyzing relative features is put forward respectively in day and night so as to optimizing feature extraction.The method of weighting feature value can improve the efficiency of pest identification. This study develops an integrative visual hardware system, including the precision baiting system of fish farming to meet the application requirements in agriculture field and the automatic detection system of agricultural pests. These hardware systems have been widely applied to demonstration garden of agricultural science and technology. In particular, precision feeding system has been applied to a fishery in Tianjing Province and pest detection system has been settled in a famous demonstration garden of agricultural science and technology in Anhui Province.