Progress in agriculture and plant breeding requires increasingly precise, fast and reproducible methods for monitoring plant traits. Plant phenotyping as a tool for quantitative assessment of morphological, physiological and biochemical traits, represents a key link between genotype and phenotype. Conventional phenotyping methods are time consuming and limiting, which has driven the development of modern, predominantly non-destructive techniques. Due to the significantly reduced measurement time and data processing requirements, these are referred to as high-throughput methods. In most cases these are imaging techniques in various spectral ranges, including RGB, infrared (IR), multispectral and hyperspectral imaging, LiDAR and chlorophyll fluorescence measurements. These systems are supported by advanced image analysis techniques. This thesis provides a systematic overview of the latest phenotyping approaches, including the underlying physical principles. It presents different platforms for data acquisition, ranging from fixed and mobile ground-based systems to multisensor airborne units, as well as their practical applications, and the role of artificial intelligence in data processing. It also highlights current challenges related to managing large data sets, standardising protocols and improving the accessibility of technologies for wider use.
|
