牧草图像快速分割方法及比较分析

THE FAST SEGMENTATION AND COMPARATIVE ANALYSIS OF PASTURE IMAGE

针对北方典型牧草植物-苜蓿图像,采用多种常用的灰度化方法,并对其中的超绿法和Cr色彩法进行了改进,结合灰度图的特点,选用了单阀值分割和基于欧式距离的聚类分割方法将图像二值化,实现了苜蓿与背景的分离提取。利用大量的图像样本,对图像的分割准确率进行了统计分析和比较研究,得到了常用算法的定量评价和分割效果,得到了针对苜蓿图像的有效分割方法,为进一步实现牧草自动识别提供了依据。

Aim at the north typical pasture - alfala image, some methods that can color - to - gray image were used to it, and improve the excess green method and Cr color method. Combined with the gray - scale image features, take the single - threshold segmentation and clustering segmentation based on euclidean distance to reach image binarization, accomplish extraction of alfalfa and background. h lot of image sapmles were used to count the segmentation accuracy, the results show quantitative evaluation and segmentation effect of commonly used algorithms, get the effective segment method to Alfalfa image, provide the basis of automatic pasture identification.