Plant Leaf Image Reconstruction Based on Point Cloud Characteristics

Abstract

We present a method of leaf image reconstruction for plants whose midribs are close to straight. First, tangerine leaves are used as an example to illustrate the geometric and leaf-shape features of such leaves and suggest an extraction method using principal component analysis. Hence, this method was used to extract the geometric features, and additional methods such as coordinate conversion, mathematical morphology, and scan line analysis were applied to extract the leaf shape features. Then, based on these leaf shape features, a Bezier deformation function of the tangerine leaf edges and their curved Bezier deformation function in the normal vector direction were employed to obtain a parametric equation of the leaves according to their rectangular deformation. Finally, a simple illumination model for the leaves is presented. The reconstruction results show that the method is highly efficient and geometrically realistic. These results will be useful for the further study of plant leaf image reconstruction.