Research And Realization Of Estimation System Of Grassland Vegetation Coverage

Vegetation coverage has always been an important criterion for vegetation considerations in the ecological field,reflecting the growth status of plants in the environment.The current coverage estimation methods mainly include sampling method,manual measurement method and algorithm measurement method.The sampling method is mainly used in images and ground measurements,sampling in a moderate proportion of actual conditions or according to specific rules,and estimating coverage by calculating the number of samples.The manual measurement method is used to calculate the area of a small range and can be used for accurate measurement.Algorithmic measurement is used in image information,such as hyperspectral images,remote sensing images,and digital images.By analyzing the inherent laws of image information,the vegetation coverage is estimated.This article uses grassland vegetation images as the research object,and uses two different methods of measurement to estimate the coverage.The main contents are as follows:1)Collecting grassland image,preprocessing and evaluation methods.We using the camera to shoot grassland images.We perform image filtering,type conversion and other preprocessing on the image,due to the presence of noise and other factors in the image,that does incompatible the experimental standards.Introducing two evaluation standards of OA and Kappa coefficient for judging accuracy.2)Grassland image segmentation based on the semi-supervised k-means algorithm of density cluster centers.Compared with the traditional k-means algorithm,the thesis calculates clusters with high density values through density clustering,takes the cluster center of the cluster as the initial cluster center,and selects the semi-supervised set of two categories(foreground,background)from the image as the constraint condition,Do k-means clustering on the image.Finally,the image is segmented according to the mapping relationship between the semi-supervised set and the cluster,and the vegetation coverage image is obtained.Experiments have proved that compared with the vegetation coverage image segmented by k-means,the segmentation effect is better and the accuracy is improved.3)Estimation of grassland coverage based on the fusion of k-means and ant colony algorithm.Aiming at the shortcomings of small feature segmentation errors in the k-means two segmentation of grass images,a grass segmentation method based on the fusion of k-means and ant colony algorithm is proposed.Using the ant colony algorithm in the image that is sensitive to the fluctuation of the neighborhood range,the characteristic contour information of the image is obtained,and the characteristic information is processed by threshold and morphology,and then merged with the k-means segmentation image.The fusion of k-means and ant colony algorithm makes up for the shortcomings of k-means in the two segmentation of grass images.Experiments show that the fusion of k-means and ant colony algorithm is better than k-means algorithm,and the effect of two segmentation in vegetation coverage image is improved.4)The system of estimation of grassland vegetation coverage.The system mainly includes that preprocessing,common image segmentation,morphological processing,the semi-supervised k-means grassland coverage estimation method based on density cluster centers,the grassland coverage estimation method based on the fusion of k-means and ant colony algorithm,etc.