Application of Image Processing Techniques in Rice Grain Phenotypic Analysis and Genome-Wide Association Studies

Background: Rice grain morphology—including traits such as awn length, hull color, size, and shape—is of central importance to yield, quality, and domestication, yet comprehensive quantification at scale has remained challenging. A promising solution has been provided by the integration of high-throughput imaging with genomic analysis. Methods: A standardized 2D image-processing pipeline was established to extract four categories of traits—awn length, hull color, projected grain area, and shape descriptors via PCA of normalized contours—from high-resolution photographs of 229 Oryza sativa japonica landraces. Genome-wide association analyses were then performed using a mixed linear model to control for population structure and kinship. Results: Broad phenotypic diversity was evident in awn length, hull coloration, grain dimensions, and morphological shape, with the first principal component explaining the dominant axis of shape variation. Known awn regulators GAD1/OsRAE2 (chr 8; ) and An-1 (chr 4; ) were identified. The hull color gene Rd (chr 1; ) was detected. A novel locus on chr 12 at 8.75 Mb with Os12g0257600 (), and the known grain size gene FLO2 (chr 4; ) were associated with projected area. Shape PC1 was mapped to GLW7/OsSPL13 (chr 7; ), NAL2/OsWOX3A (chr 11; ), and OsGIF1 (chr 11; ). Conclusions: This study demonstrates that image-based phenotyping combined with genome-wide association studies (GWAS) can efficiently reveal both established and novel genetic determinants of rice grain morphology. These findings provide actionable targets for marker-assisted selection and genome editing to tailor grain traits in rice breeding programs.