摘要
单位面积麦穗数是小麦产量预测的一个重要参数,如何从图像上自动识别出麦穗数是测产的关键。为此,使用Sobel算子对麦穗图像进行边缘检测,使麦穗从混有少量杂草的模糊的背景中分割开,并与加权平均法、G分量法和最大值法处理后的图像进行了比较。随机选取麦穗无交叉的50幅图像样本,分别使用上述方法处理,Sobel算子法与其他3种方法相比,图像分割的总体耗时至少减少了10%。实验结果表明,Sobel算子对麦穗图像分割是有效的。
The number of wheat panicle is an important parameter in the wheat yield forecast. How to identify the number of wheat from the image is the key of yield monitor. The Sobel operator edge detection image was used image segmentation of wheat panicle. Panicle were separated from the blurred and mixture a small amount of weed of background. Then com- pared this method with the weighted average method, the G component method and the maximum value method, the out- line of panicle were much clearer and more explicit than other three methods. Randomly selected 50 independent and o- verlapping panicle images as samples, using the above approach for processing, the segmentation based on Sobel operator of the overall time-consuming compared with the other three methods were increased by at least 10%. The experimental results show that the Sobel operator on the wheat image segmentation is effective.
出处
《农机化研究》
北大核心
2013年第3期33-36,共4页
Journal of Agricultural Mechanization Research
基金
国家自然科学基金项目(60873236)
河北省教育厅科学研究项目(2010251)
河北省教育厅科学研究项目(Z2009122)
关键词
麦穗
SOBEL算子
图像分割
wheat panicle
sobel operator
image segmentation
