Panjvani Karim, Dinh Anh V, Wahid Khan A
Department of Electrical and Computer Engineering, University of Saskatchewan, Saskatoon, SK, Canada.
Front Plant Sci. 2019 Feb 13;10:147. doi: 10.3389/fpls.2019.00147. eCollection 2019.
The ever-growing world population brings the challenge for food security in the current world. The gene modification tools have opened a new era for fast-paced research on new crop identification and development. However, the bottleneck in the plant phenotyping technology restricts the alignment in geno-pheno development as phenotyping is the key for the identification of potential crop for improved yield and resistance to the changing environment. Various attempts to making the plant phenotyping a "high-throughput" have been made while utilizing the existing sensors and technology. However, the demand for 'good' phenotypic information for linkage to the genome in understanding the gene-environment interactions is still a bottleneck in the plant phenotyping technologies. Moreover, the available technologies and instruments are inaccessible, expensive, and sometimes bulky. This work attempts to address some of the critical problems, such as exploration and development of a low-cost LiDAR-based platform for phenotyping the plants in-lab and in-field. A low-cost LiDAR-based system design, LiDARPheno, is introduced in this work to assess the feasibility of the inexpensive LiDAR sensor in the leaf trait (length, width, and area) extraction. A detailed design of the LiDARPheno, based on low-cost and off-the-shelf components and modules, is presented. Moreover, the design of the firmware to control the hardware setup of the system and the user-level python-based script for data acquisition is proposed. The software part of the system utilizes the publicly available libraries and Application Programming Interfaces (APIs), making it easy to implement the system by a non-technical user. The LiDAR data analysis methods are presented, and algorithms for processing the data and extracting the leaf traits are developed. The processing includes conversion, cleaning/filtering, segmentation and trait extraction from the LiDAR data. Experiments on indoor plants and canola plants were performed for the development and validation of the methods for estimation of the leaf traits. The results of the LiDARPheno based trait extraction are compared with the SICK LMS400 (a commercial 2D LiDAR) to assess the performance of the developed system.
不断增长的世界人口给当前世界的粮食安全带来了挑战。基因编辑工具为新作物鉴定与培育的快速研究开启了新纪元。然而，植物表型分析技术的瓶颈限制了基因-表型发展的匹配，因为表型分析是鉴定具有更高产量和适应不断变化环境能力的潜在作物的关键。人们利用现有传感器和技术，为使植物表型分析实现“高通量”做出了各种尝试。然而，在理解基因-环境相互作用时，将“优质”表型信息与基因组相联系的需求仍是植物表型分析技术的一个瓶颈。此外，现有的技术和仪器难以获取、价格昂贵，且有时体积庞大。这项工作试图解决一些关键问题，比如探索和开发一个低成本的基于激光雷达的平台，用于在实验室和田间对植物进行表型分析。本文介绍了一种基于低成本激光雷达的系统设计LiDARPheno，以评估廉价激光雷达传感器在叶片特征（长度、宽度和面积）提取中的可行性。文中给出了基于低成本和现成组件与模块的LiDARPheno的详细设计。此外，还提出了用于控制系统硬件设置的固件设计以及用于数据采集的基于用户级Python的脚本。该系统的软件部分利用了公开可用的库和应用程序编程接口（API），使得非技术用户也易于实现该系统。文中介绍了激光雷达数据分析方法，并开发了用于处理数据和提取叶片特征的算法。处理过程包括激光雷达数据的转换、清理/过滤、分割和特征提取。针对室内植物和油菜进行了实验，以开发和验证叶片特征估计方法。将基于LiDARPheno的特征提取结果与SICK LMS400（一种商用二维激光雷达）进行比较，以评估所开发系统的性能。
