基于SFM+光学校正的小麦根系三维重建与表型提取方法

Three Dimensional Reconstruction and Phenotype Extraction of Wheat Root System Based on SFM and Optical Correction

摘要:

小麦根系作为关键的水分和养分吸收器官，其表型特征对了解小麦生长状况及土壤环境具有重要意义。然而，根系地下生长的特性使得其观测面临挑战。本研究设计了一套小麦水生培养装置、基于虹吸原理的柔性换液装置及根系图像采集系统，并针对该系统开发了图像光学折射校正方法，通过SFM算法构建了根系的三维点云模型，并提取了相关表型特征。试验表明，柔性换液装置使换液前后根系图像结构相似度提高至0.98；折射校正方法使图像误差减少62%。采用提出的装置及方法对小麦根系生长发育受氮环境的影响进行了探讨。研究结果表明，在低氮条件下，小麦根系展现出更深、分布更密集的生长趋势。另外，相比于氮高效品种，氮低效品种对氮环境变化更敏感。本文提出的装置及方法有助于高通量植物根系三维表型分析。

Abstract:

Wheat is one of the most important food crops in the world, and its root system serves as a key organ for water and nutrient absorption. Its phenotypic characteristics are of great significance for understanding the growth status of wheat and soil environment. However, the underground growth characteristics of the root system pose challenges for its observation. A wheat aquatic cultivation device, a flexible liquid exchange device based on siphon principle, and a root image acquisition system were developed. An image refraction correction method was developed for the system. Then a three-dimensional point cloud model of the root system was constructed by using SFM algorithm, and relevant surface features were extracted. Experimental results showed that the flexible liquid exchange device increased the similarity of root image structure before and after liquid exchange to 0.98. The refraction correction method reduced image errors by 62%. Furthermore, the proposed device and method were utilized to investigate the influence of nitrogen environment on the growth and development of wheat roots. The research results indicated that under low nitrogen conditions, wheat roots exhibited a deeper and more densely distributed growth trend. In addition, compared with nitrogen efficient varieties, nitrogen inefficient varieties were more sensitive to changes in nitrogen environment. The device and method proposed can contribute to high-throughput three-dimensional phenotype analysis of plant roots, opening up a perspective for root research in the development of smart agriculture driven by big data.