摘要
碎米率是衡量大米的外观品质的一个重要指标。在使用机器视觉技术自动检测碎米率的过程中,大米籽粒粘连的情况常常不可避免。为了分割出单颗大米,提出了一种改进的、基于形态学重建处理的标记分水岭算法,可在分割出的单粒大米的基础上,确定大米的数量,并利用外接矩形来确定米粒的长度,从而检测出碎米率。实验结果表明,在大米粘连程度及含有的碎米量不同的情况下,此方法都具有比较高的检测精度。
Rate of broken rice is an important indicator in the measuring of appearance quality of rice. In the process of automatic detection on machine vision technology, touching rice grains are usual. To get every single grain, an improved marker watershed algorithm based on morphology was proposed. Every single grain can be segmented, so the numbers of rice grains and the length of a rice grain can be obtained. According to the results, the rate of broken rice in the samples can be accurately determined. The experiments show that the method is efficient and effective in the state of different rate of broken rice and different adhering extent between the rice grains.
出处
《农机化研究》
北大核心
2015年第2期222-225,共4页
Journal of Agricultural Mechanization Research
基金
贵州大学青年基金项目(贵大自青合字2012014)
贵州省自然科学基金项目(2011-2103)
国家自然科学基金项目(11264006)
关键词
粘连大米
碎米率
分水岭算法
检测
定标
adhering rice grain
scale
broken rice rate
watershed algorithm
detection