Research On Bursaphelenchus Xylophilus Area Detection Based On Remote Sensing Image

Bursaphelenchus xylophilus is one of the most dangerous forest pest, the pine which infected with it would be exhausted to death. The disease originated in North America, and many countries all have happened,too. In our country,it was first discovered in1982. in just more than ten years, the disease successively happened in Jiangsu, Anhui, Shandong, Zhejiang, Guangdong, Hubei, Hunan, Taiwan, Hong Kong and many other areas. The onset of the disease from hundreds of hectares scale to tens of thousands of hectares, killing a large number of pine trees. Bursaphelenchus xylophilus Nickle’s spread will bring a serious influence on our country’s forest resources and forestry ecological civilization construction, not only bring huge economic losses to the production of forestry, but also ruined the forestry ecological construction seriously.Due to serious damage of the Bursaphelenchus xylophilus Nickle and the difficult prevention for it, many scholars have studied for it. They also proposed some monitoring methods, but these monitoring method often has strong subjectivity, or difficult to implementation, Bursaphelenchus xylophilus Nickle cannot be found in a timely and a accurate manner, making timely unable to make effective prevention and control measures. This article uses the Unmanned Aerial Vehicle(UAV) technology through the UAV to collect images of forest terrain information, using computer technology processing the collected information, so as to realize the recognition and monitoring to the infected trees, and according this to positioning the infected area, prevention and control this disease, to achieve the purpose of protecting pines. It has very vital significance to our country’s forestry economy and forestry ecological construction.In fact,the detection of plant diseases and insect pests areas can be extended to feature extraction and classification of the UAV remote sensing image. This article uses the UAV aerial technology through the UAV aerial image to collect forest terrain information, starting from the goal of UAV remote sensing image classification, study of plant diseases and insect pests detection in remote sensing images, the main content includes the following aspects:1、Introduces the basic theory used in pest detection which based on remote sensing images. Including:discusse the UAV aerial remote sensing image’s feature extraction method; Introduce some commonly used classification algorithms and analyzes their advantages and disadvantages and applicable situations; Introduction and analysis of several kinds of interested region detection method, etc.2、Depth of the neural network is introduced. Mainly include the depth neural network training step by step in the process of greed training and supervision of the master, analysis its advantages compared with other classification algorithms.3、Put forward a plant diseases and insect pests detection method which based on a depth neural network classification algorithm. The algorithm was improved and extend from the artificial neural network, is a kind of approximate algorithm of computer. Each pixel in this method based on remote sensing image to meet the need of feature extraction for classification of feature vector, and using the depth neural network classification methods on the of these feature vector to classify them, implementation of remote sensing image in the area of plant diseases and insect pests for testing purposes. By experimental verification, this method can more accurate detection the plant diseases and insect pests area.4、Put forward a plant diseases and insect pests detection method which combination interested region detection and texture feature detection. Based on frequency domain by using the method to adjust the interest area detection, extraction from remote sensing images, including pest region, interested in all areas, and in the depth of interested in these area, using the neural network classification methods to classify these regions. Through experimental verification, the method can achieve better detection of plant diseases and insect pests areas, and reduces the computational complexity greatly.