Codelplant: Regression-based processing of RGB images for colour models in plant image segmentation

摘要

Analysis of colour in digital images has recently gained prominence in plant studies as a result of the reduction in the cost of imaging devices, the increased resolution of images and the capacity for processing and analysis. We present Codelplant, a generic open-source code that can be used to generate models characterising subset-specific colours in digital images. The method is based on the creation of a statistical regression-based model from data obtained by analysis of RGB pixels. As the model is supervised by human observation, the human factor is included in the process. The method and code can be used as the basis of other approaches (e.g. in agriculture or forestry) or for further development.