Development of an Automated High- Throughput Phenotyping System for Wheat Evaluation in a Controlled Environment

Plant breeding has significantly improved in recent years; however, phenotyping remains a bottleneck, as the process of evaluating and measuring plant traits is often expensive, subjective, and laborious. Although commercial phenotyping systems are available, factors like cost, space, and need for specific controlled-environment conditions limit the affordability of these products. An accurate, user-friendly, adaptive, and high-throughput phenotyping (HTP) system is highly desirable to plant breeders, physiologists, and agronomists. To solve this problem, an automated HTP system and image processing algorithms were developed and tested in this study. The automated platform was an integration of an aluminum framework (including movement and control components), three cameras, and a laptop computer. A control program was developed using LabVIEW to manage operation of the system frame and sensors as a single-unit automated HTP system. Image processing algorithms were developed in MATLAB for high-throughput analysis of images acquired by the system to estimate phenotypes and traits associated with tested plants. The phenotypes extracted were color/spectral, texture, temperature, morphology, and greenness features on a temporal scale. Using two wheat lines with known heat tolerance, the functions of the HTP system were validated. Heat stress tolerance experiments revealed that features such as green leaf area and green normalized difference vegetation index derived from our system showed differences between the control and heat stress treatments, as well as between heat-tolerant and susceptible wheat lines. In another experiment, stripe rust resistance in wheat was assessed. With the HTP system, some potential for detecting qualitative traits, such as disease resistance, was observed, although further validation is needed. In summary, successful development and implementation of an automated system with custom image processing algorithms for HTP in wheat was achieved. Improvement of such systems would further help plant breeders, physiologists, and agronomists to phenotype crops in an efficient, objective, and high-throughput manner. Keywords: Automation, Heat stress, Image processing, Plant breeding, Sensing, Stripe rust.