Recognition Of Plants With Complicated Background By Leaf Features

Plant taxonomy is an important basic subject,and its classification research helps people to understand,utilize and protect plants.With the development of computer technology and machine vision,plant classification recognition based on computer vision has become a hot topic in the research of plants taxonomy.As an important part of plants,leaves are usually used as an important target of plant recognition.At present,most researches on plant classification and recognition are with a single background.Complicated background can't be avoided in the process of leaves image extraction,and the research of leaves recognition based on complicated background is less.In this paper,plant leaves recognition is studied under complicated background.Plant recognition with complicated background is mostly based on segmentation and recognition.The existing research of segmentation has poor effect and prone to over segmentation.Aiming at the existing problems in the research of plant recognition with complicated background,In this paper,plant leaf segmentation algorithm and leaf feature extraction algorithm are studied respectively.A recognition algorithm of plant leaves with complicated background is proposed.he main contents of this paper are as follows:1.Plant leaves segmentation: Research on the maker watershed algorithm.According to the characteristics of leaves with complicated background,the leaf image is transformed to super green space,remove the non green background by threshold segmentation.Then transform leaf image from RGB space to HSI space.Morphological operation is used to reconstruct leaf image with complicated background based on opening and closing.Then the reconstructed image is segmented by maker watershed segmentation2.Leaf features extraction: After segmentation,the single background leaves is preprocessed,and then the block LBP is extracted as the leaf texture feature,the Hu moment invariant is extracted as the shape feature.3.Dimensionality reduction of Leaves feature: In order to improve the efficiency of features recognition,we use LLE(local linear embedding)algorithm to reduce the dimension of block LBP(Local Binary Patterns)features.4.Plant leaves classification recognition: On the basis of the above research,plant leaves were collected to build leaf bank,and SVM classification and recognition system was built on MATLAB platform to classify and identify leaf features.The results of leaves segmentation show that this algorithm can effectively segment various leaves in complex background,and the leaves classification and recognition algorithm based on feature dimension reduction and fusion can effectively improve the efficiency of leaves recognition.