摘要：Tiller, as an important agronomy trait, determines panicle number contributing to grain yield. For replacing traditional inefficient methods of counting tillers, rapid and accurate measurement of pheno-type in modern breeding has become urgent recently. Therefore, to automate its process and improve its accuracy and efficiency, many high-throughput facilities were developed, some of which base on conventional X-ray computed tomography (CT) system. However, CT system should be placed in a security apartment, in case of exposing operators into radiation and causing experiment objects gene mutation. In this study, initially, magnetic resonance imaging which features non-destructive and non-radiation is applied to obtain transversal section images of rice culms. After that, these MRI images were processed by separating algorithm which can extract tillers’ section and acquire its number based on morphological parameters. Finally, by comparing the processed image of MRI with CT’s, it proves that this algorithm can capture clear segmented image as CT does. Meaningfully, it turns out that this method can be recommended as an efficient replaceable way for studying rice tiller’s agronomy in breeding work.
