Comparison And Pattern Dynamics Of Alien Invasive Plant Extraction Methods Based On Worldview Images

With the deepening of the process of global integration,the invasion of alien species is one of the important ecological problems brought about by global integration.The rapid progress of urbanization in China has also made the impact of human activities on ecosystems more and more frequent,and the opportunities for invasive alien species have become more and more.The invasion of alien species poses a great threat to the national economy,public health,agricultural productivity and ecological integrity.The spread of invasive alien species is extremely fast.It is more feasible and economical to predict the spread of the invasive species according to the growth and survival range.The study of reproducible monitoring tools is also very necessary.Taking the area around the estuary of Jitimen waterway in Zhuhai City,Guangdong Province as the research area and the water hyacinth as the research object,this paper takes Worldview-3 high resolution remote sensing image and field survey content as the data source,selects the commonly used alien invasive plant classification method to extract the growth area of water hoist in the study area,combined with visual interpretation and classification accuracy evaluation.The applicability of each method in extracting water hyacinth from high resolution remote sensing images is compared and analyzed,and the changes of water hyacinth growth area and the change of distribution pattern in 2014 and 2017 are analyzed,and the following conclusions are obtained:(1)The object-based nearest classification method is the best for the extraction of water hyacinth.The K-means method and the ISODATA method cannot separate the water hyacinth from other vegetations;the support vector machine method for extracting the water hyacinth growth area has a user accuracy of 71% and a charting accuracy of 62.5%;The user accuracy and charting accuracy of the maximum likelihood method classification result are 86% and 78.8%.The classification results have a certain phenomenon of salt and pepper,and the classification is too fragmented;In the results of threshold classification for the extraction accuracy of the growth region of water hyacinth,the user accuracy is 74%,and the drawing accuracy is 87.5%.Compared with the leakage phenomenon,the misclassification phenomenon is more serious;The nearest neighbor classification method has a user accuracy of 84% and a charting accuracy of 96.3% when extracting the water hyacinth growth area.(2)In the nearest neighbor classification and threshold classification,the most suitable segmentation parameters for extracting water hyacinth are: segmentation scale as 100,spectral factor as 0.6,shape factor as 0.4,tightness factor as 0.5,smoothness factor as 0.5.Classification features include spectral characteristics of 8 bands,NDVI,NDWI.The entropy of the green band and the mean of the near-infrared 1 band in the texture feature are also used as classification features to distinguish between water hyacinth and other vegetation types.(3)The water hyacinth growth area is increasing and is in a more highly aggregated state.Most of the water hyacinth distribution areas are in rivers with a width of less than 20 m.Compared with 2014,the growth area of water hyacinth in the river channel with a width greater than 20 m in 2017 is reduced,and the growth area of water hyacinth in the river channel with a width less than 20 m is increased.This paper studies and compares the methods of obtaining alien invasive species based on high-resolution remote sensing images,and provides some reliable reference for later related research.Applying scientific methods to rationally analyze ecological problems in reality,and provide a scientific basis for the government to formulate relevant policies and management plans.