Comparison Of Segmentation Methods For Achnatherum Splendids Community Using Uva Remote Sensing Images

With the gradual increase of human activities and the deterioration of the natural environment,the ecological environment of Hulun Buir Grassland is facing a great threat.Due to the special geographical location and climate in Hulun Buir Grassland,the growth of a large number of achnatherum splendid(A.splendens)community community,other rare herbs.The vegetation coverage is generally regarded as an important index of grassland desertification evaluation and land desertification assessment,vegetation coverage of Hulun Buir Grassland splendens community is conducive to the accurate estimation,soil erosion,gas exchange and understanding,can be a good indication of climate change and ecological environment change.This paper takes the typical A.splendens community in Hulun Buir Grassland as the research object,the UAV remote sensing images and image segmentation technology based on the estimation of A.splendens community vegetation coverage by the Otsu algorithm and IAFSA-KM algorithm,the overall estimated by part of the overall ecological environment in Hulun Buir Grassland in a quantitative and qualitative assessment.The scheme mainly includes 3 steps:image preprocessing,maximum inter class variance algorithm,image segmentation of IAFSA-KM algorithm,and contrast analysis of experimental results.The main research contents are as follows:1.Data preprocessing based on unmanned aerial vehicle(UAV)remote sensing image.First,we made a general understanding of the research area and analyzed and understood the data collected by Phantom4,and then processed the data preprocessing of image clipping,geometric correction and Gamma correction.The image after data preprocessing does not change the basic information of the image,but it will provide a great convenience for the subsequent image segmentation.2.Based on the image segmentation technique,the threshold segmentation of the maximum inter class variance algorithm is proposed.First,to analyze the image,enhance the information on the boundary between the a.splendens community and non splendens using gradient transform and gray transform,achieves good experimental results;because the CMOS sensor UAV using the aerial process comes with some noise,noise analysis categories,describes the principle and process of adaptive median filtering algorithm and using adaptive median filtering algorithm to filter the image noise;then using Otsu threshold segmentation to find the best treatment of images;finally,using the image processing method of mathematical morphology,optimize the segmentation boundary information.3.Based on the image segmentation technique,the image segmentation of IAFSA-KM clustering algorithm is proposed.First,we analyze the clustering principle of K-means clustering algorithm and the results of image segmentation,and find that the segmentation results do not achieve the desired results.In view of the sensitivity of the K-means clustering algorithm to the selection of the initial cluster center,it is easy to fall into local convergence.In this experiment,the artificial fish swarm algorithm to K-means clustering algorithm,the combination of the IAFSA-KM algorithm,the algorithm based on artificial fish swarm algorithm to reduce the iterative process of K-means clustering algorithm,the optimized selection of initial point,so as to achieve the global optimal clustering,K-means clustering algorithm to improve the stability,optimize the image segmentation results.4.The comparison and analysis of the experimental results.In the last section of the third chapter and the fourth chapter makes a comparison and analysis of experimental data,the vegetation coverage as the accuracy measure,the use of foreign professional treatment of vegetation coverage of the caneye software as compared experimental results accuracy,the maximum variance algorithm in the third chapter,the fourth chapter(Experiment one)and experimental(two)the experimental results show that the vegetation coverage of the IAFSA-KM algorithm,MSE analysis and comparison of four kinds of methods and results of regression comparison,thus obtains the IAFSA-KM optimal algorithm.The accuracy of the experimental results showed that the experiment(two)>experiment(one)>caneye vegetation coverage.The superiority and feasibility of the improved IAFSA-KM clustering algorithm are proved.