Design And Implementation Of Botany Feature Extraction Methods Based On Leaf Shape

Plants serve as an integral part of the earth’s biosphere. Mankind has to recognize the plants before developing and utilizing it. However, it is a skill that could only be possessed through long-term learning and accumulation as the result of the varieties of plants. Currently, there is a great scarcity of talents in the field of classical plant taxonomy. In this sense, it is of paramount importance to both teaching and the science popularization education of the general public to develop a practical and rapid identification technology of plants; the key to which lies in the feature extraction of various plants.Recent years have seen the research on the extraction of the features of the plants leaves by means of computer. The plants varieties are finally identified through the extraction and analysis of leaf shape, vein, texture or the features of the image frequency domain of leafs. This thesis offers an investigation of the progress and current status of the extraction of the image feature of plant leaves and classifies the leaf shapes into four categories, namely, the overall leaf shape, leaf apex, leaf base and leaf margin on the basis of the knowledge about the classical plant taxonomy with different characteristic parameter given to each category. At last, it realizes the image preprocessing of the plant leaf blade samples, image segmentation and features extraction through the OpenCV.According to the basic thoughts of the methods introduced in this thesis, the leaf shapes are subdivided and the local characteristic parameter extracted according to the principle consistent with the plant taxonomy, which is different through the previous thoughts of extracting the overall characteristics parameters of the leaf blade image and lays a foundation for the progressive and hierarchical identification method. This thesis implementation the botany feature extraction Methods by OpenCV, thus providing the reference for the rapid establishment of an experimental platform for leaf blade image processing and feature extraction. Besides, this paper also offers an introduction to the correct method of collecting the leaf blade sample as well as the method of obtaining the clear image of leaf blade.