基于改进K-means算法的冬小麦覆盖度提取研究

Extraction of Winter Wheat Coverage Based on Improved K-means Algorithm

为快速、精准地提取冬前分蘖期冬小麦覆盖度,提出了一种基于改进K-means算法的冬小麦覆盖度提取方法。首先将冬小麦图像转换到Lab色彩空间,其次利用蜉蝣算法(Mayfly Algorithm, MA)获取K-means最优初始聚类中心,并用马氏距离代替欧氏距离进行算法改进,最后利用分割得到的二值图像计算冬小麦覆盖度。结果显示,该方法的平均分割精度和平均处理时间分别为94.66%和2.03 s,与过绿指数(excess green,EXG)自适应阈值分割和基于粒子群优化算法(particle swarm optimization,PSO)的K-means(PSO-K-means)分割相比,分割精度分别提高了12.04%和4.18%,处理时间分别减少了2.26和2.94 s。该方法分割效果优于EXG和PSO-K-means分割方法,可用于提取冬小麦覆盖度。

In order to rapidly and accurately extract winter wheat coverage at tillering stage before winter,a new method based on the improved K-means Algorithm was proposed.Firstly,the image of winter wheat was transformed to Lab color space.Secondly,the initial clustering center of the K-means Algorithm was obtained by using the Mayfly Algorithm(MA).At the same time,the Algorithm was improved by using Mahalanobis distance instead of euclidean distance.Finally,the coverage of winter wheat was calculated by using the segmented binary image.The test results of 100 winter wheat images showed that the average segmentation accuracy and the average processing time of the proposed method were 94.66%and 2.03 s,respectively.Compared with excess green(EXG)adaptive threshold segmentation and K-means segmentation based on particle swarm optimization algorithm(PSO-K-means),the average segmentation accuracy increased by 12.04%and 4.18%,respectively.And the average processing time was reduced by 2.26 and 2.94 s,respectively.The results showed that the segmentation effect of this method was better than the other two segmentation methods,and it could be used to extract the coverage of winter wheat.