Realistic image synthesis of plant structures for genetic analysis

摘要

The objective of this study was to develop a systematic method to restore images of plants using a small number of parameters in order to facilitate investigation of genetic behaviour of branching patterns by computer models. Based on biological data, a generic methodology for image synthesis of plant structures was proposed. The corresponding algorithms for feature extracting and tuning were designed. Images of plant structures were synthesised with a higher accuracy by a model that was constructed using parametric L-systems. The algorithms were validated by comparing the synthetic images with the real plants. The approach makes it possible to assess the virtual architectures of plants by computer models, instead of real experiments, and to investigate the relationship between plant morphology and their genetic behaviours. This may aid the understanding of complex branching patterns that result from a plant's genetic behaviour.