Remote Sensing Extraction of Crop Disaster Information Based on Support Vector Machine
Crop pests and diseases are the most terrible disasters in agricultural production. First, crop pests and diseases account for a large proportion of crop disasters. Secondly, it is the main reason that restricts the production of high efficiency, high quality and high yield of crops. As a large agricultural country, crop pests and diseases occur in a wide variety and have a wide range of impacts, causing huge losses to China's grain production. In order to provide a basis for reasonable control of crop pests and diseases, it is necessary to timely identify the actual disaster situation of crops. In the traditional crop damage detection, due to the backwardness of technology, people have to use the method of visual observation to detect and occur the specific situation of crop pests and diseases and to judge the way of the outbreak of pests and diseases. The traditional method is not only a waste of time but also not very efficient, and the monitoring effect is not high. Therefore, in order to effectively increase the level of monitoring and accuracy, it is necessary to make rational use of hyperspectral remote sensing technology. In this paper, the common pests and diseases of rice are taken as the research object. The monitoring on the regional scale is the main line of research. The remote sensing satellite image data and the environmental and disaster monitoring and forecasting small satellite images are used as data sets to study the remote sensing monitoring models and methods of rice pests and diseases. In this paper, a method of Remote Sensing Extraction of crop disaster information based on support vector machine is proposed. Firstly, according to the characteristics of different regions, each color component in visible light remote sensing image and near-infrared remote sensing image is extracted as the color feature of the corresponding region, and then the windowed image is added. The gray level co-occurrence matrix is extracted as the texture feature of the central pixel. Finally, the training samples and the support vector machine training model are established to extract the information of rice pest diseases.