The Research Of Similarity Comparising Method Of Plant Shapes

The difference of plant external morphological is often considered as the key factors in the work of plant classification and identification. It has different morphology, structure and habits between different kinds of plants. These differences can be used to determine the closeness of their genetic relationship. The research of plant morphological similarity calculated method can provide new methods and tools for plant taxonomy. Furthermore, the key question to judge the accuracy of the simulation model in the field of virtual plant is that judging the similarity between the reconstructed3D models and real plants. Therefore, the research of plant morphological similarity comparison method has important theoretical significance and practical value in many fields, such as plant classification, plant identification, plant retrieval, virtual plants.Literature researches find that the similarity in the three-dimensional model of research mainly concentrates on the mechanical CAD/CAM, the biological field (the structure of DNA and protein sequences), network security (malicious code similarity comparison). However plant morphology similarity calculation is relatively rare, and very weak. In response to this situation, this paper discusses about plant morphological similarity calculation method, which is based on tree diagram similarity and graphics similar elementary theory. The main work is as follows:1) Based on plant morphology similarity definition, we proposed the plant overall morphology similarity calculation method which made from the three aspects of the topological structure of the plant, the appearance of contour and internal minutiae.2) Proposed plant topology similarity calculation method which based on a simplified tree diagram, edit distance and node mapping constraints.3) Proposed a new plant external contour similarity calculation method. This paper presents a two-dimensional graphics similarity measure algorithm, and extended to three-dimension, applied to the tree structure and its sub trees’contour similarity measure.4) Proposed a new similarity measurement method between plants internal detail features. It’s considered from the average axial angle, the diameter ratio between rank one lateral branch and trunk, the overall width-height ratio, the average axial angel of rank one and rank two lateral branches, the cross-sectional area ratio between rank one lateral branch and trunk, and so on.5) Taking advantage of aforementioned methods of the plant morphology similarity, plant species identification system to distinguish different kinds of plants was developed under Ubuntu Linux operating system. Experiments verified the algorithm and software system respectively from the three aspects of the topology, the appearance of the outline and internal detail features.The plant morphological similarity calculation method of this study can effectively calculate the degree of similarity between the different plants. Plant species identification system developed on this basis, can be used to discriminate the degree of coincidence between automatically generated or hand-designed tree model and real plants, and thus has some practical value in the field of plant pattern recognition.