CAJ | 학술논문

作物籽粒数常用于评价农作物的产量和品质。为此，利用机器视觉技术，提出了一种基于圆心定位的籽粒计数方法。针对不粘连、简单粘连和复杂粘连3种情况，各拍摄15张大豆照片，利用MatLab进行图像预处理、粒径统计、Freeman 链码提取、圆心定位、圆心筛选和聚类等处理后，计算大豆的籽粒数。实验表明：不粘连和简单粘连准确率均为100％，复杂粘连的平均准确率为97．8％；此方法对于圆形籽粒的计数具有很好的效果。
작물자립수상용우평개농작물적산량화품질。위차，이용궤기시각기술，제출료일충기우원심정위적자립계수방법。침대불점련、간단점련화복잡점련3충정황，각박섭15장대두조편，이용MatLab진행도상예처리、립경통계、Freeman 련마제취、원심정위、원심사선화취류등처리후，계산대두적자립수。실험표명：불점련화간단점련준학솔균위100％，복잡점련적평균준학솔위97．8％；차방법대우원형자립적계수구유흔호적효과。
The grain number of crops has been used commonly for assesment of yield and quality.Based on the machine vision technology, this study carried out a simple method of localization of centers for counting of grain particles.For three condition of no-adhesion, simple-adhesion and complex-adhesion, we took 15 pictures of soybeans respectively. After consecutive image operation of preprocessing, statistics of granulometry, extraction of Freeman code, calculation of centers, sieving and clustering using Matlab IPT toolbox, we got the total number of soybeans.The obtained reasult was that, both no-adhesion and simple-adhesion got 100%accuracy rate, and the complex-adhesion achieved accuracy rate 97.8%in average.The experiment proved the feasibility of this method in the counting of round like grain particles.