基于同态滤波和K均值聚类算法的草原植被盖度测量

Measuring Grassland Vegetation Cover Based on Homomorphic Filtering and K-means Clustering Algorithm

针对草原盖度地表测量法费时、费力、重现性差及自然环境下光照不均草原植被图像分割效果不理想问题,应用动态巴特沃斯同态滤波法对草原植被图像进行光照补偿,采用K均值聚类算法对补偿后图像进行分割,最后根据植被盖度定义,实现草原植被盖度测量。试验结果表明:本算法测量标准差、相对误差、均方根误差RMSE和耗时均值分别为0.5 7 0%、3.9 8 8%、0.1 0 0和2.3 5 s,比方格纸测量法标准差低4.6 8 6%,速度提升9 0倍左右。通过对分割效果定性和定量分析,验证了本文算法的草原植被盖度测量精度、稳定性和速度,可为草原盖度研究提供参考和技术支持。

In view of the problems such as time-consuming,laborious and reproducibility of the surface measurement method of grassland coverage,the effect of uneven vegetation image of vegetation in natural environment is not ideal. Applying dynamic butterworth homomorphic filtering method to compensate grassland vegetation images,K-means clustering algorithm is used to segment the compensated image. Finally,the vegetation coverage is measured according to the definition of vegetation coverage. The experimental results show that the standard deviation,relative error,root mean square error RMSE and time consuming mean are 0. 570%,3. 988%,0. 100 and 2. 35 s respectively,the standard grid measurement method standard deviation of 4. 686%,the speed increase of about 90 times. Based on the qualitative and quantitative analysis of the segmentation effect,the accuracy,stability and velocity of grassland vegetation coverage are verified by this algorithm,which can provide reference and technical support for grassland cover research.