摘要
以麦田为例,对不同季节的成熟和未成熟麦田场景,选择在HSI颜色模型下进行图像处理,采用改进的区域分裂聚合算法对图像进行分割,并利用掩膜过滤机制去除非道路部分连通区域,最后采用改进的Moore边界追踪算法快速准确地提取出田间道路边界。对多幅农田图像试验结果表明,该方法处理一幅640×480像素的彩色图像只需0.325 s,能够快速准确地对多种场景下田间道路边界进行检测,很好地过滤阴影、农作物不规则覆盖等干扰。
The HSI color model was employed to deal with the mature or immature wheat field scenes, and the im- proved region splitting and merging algorithm was adopted to split the images. Then the mask filter mechanism was applied to eliminate the eonnected area not belonging to the path. Finally the improved moore boundary tracing algorithm was intro- duced to find the field path boundary. The experiment results on multiple farmland images showed that the time of process- ing a 640x480 pixel color image was only 0.325 second and the method can quickly and accurately detect the boundary of farmland field path under various scenes with good filtering of shadow and irregular cover of crop.
出处
《江苏农业学报》
CSCD
北大核心
2013年第6期1291-1296,共6页
Jiangsu Journal of Agricultural Sciences
基金
国家科技支撑计划项目(2011BAK21B05)
中央高校基本科研业务专项资金项目(Y0201100080)
关键词
机器视觉
图像分割
道路检测
computer vision
image segmentation
path detection
