Study of Non-Parametric Aberration Corrections for the Regular Camera Images Used in Measuring Plant Appearance Phenotype

The residual aberrations in ordinary optical camera images are one of the main obstacles to the measurement accuracy of the shape of plant organs (such as plant leaves), so it is necessary to correct the position of the pixel. In the parametric correction algorithms, the mapping between two-dimensional images and three-dimensional objects is regarded as an approximate linear system; and its accuracy cannot meet the requirements of forward intersection measurement. Based on Gaussian imaging and perfect imaging conditions, this study analyzes the causes of aberrations in optical imaging systems, and disassembles the aberrations into vertical aberrations and axial aberrations. The vertical aberration is modeled based on Gaussian ideal imaging. According to the perfect imaging conditions, the axial aberration of a pixel is considered to be the curl of the axial aberration vectors in the neighborhood, and can be disassembled into the axial geometric aberration component and the axial wave aberration component. The forward intersection measurement results of the corner points of the target grid and the areas of the plant leaves are satisfactory, indicating that the proposed image correction method is accurate and effective.