Herb Image Feature Extraction And Classification Research

The hot spot in the recent study is that we can use the leaves’ images to achieve the identification and classification of plant species automatically. Scholars from the world are all based on the extraction and identification of the leaves’ features to carry on their researches and they have already achieved some researching results.However, according to the researching results and recent researching situation, the technology of plant identification still remains problems such as the lack of the definitions of plants’ feature and extraction of blade aspect ratio, rectangular, eccentricity. These features can’t describe blade specifically.Aiming at the problems existing in the feature extraction of plant leaves, this passage uses four kinds of herbs’leaves and comes out four new features of the saw tooth and leaf crack. Making the use of morphological erosion and dilation fills small holes and the contours of the fracture and takes a Gaussian filter to filter noise. Combining the inflection point detection method based on chain code with Chain code poor estimation method based on the boundary of the convex determine the jagged points and fissure points and calculating the chain code target area base on a Improvement way. Using statistical methods based on gray level co-occurrence matrix to extract herbaceous plants leaves’texture characteristics such as energy, entropy, contrast. According to the extracted Eigen value of herbaceous plant leaves and base on the supporting vector machine classifier for herbs, identification and classification of the experiment show that the classifier can get higher recognition rate and have good recognition effect.The new proposed method which mentioned in thus message can greatly improve the accuracy of the calculation of the area, and the proposed new way of identification about plant species according to the Eigen values which riches the mathematical theory based on the identification of plant species of leaf traits.The study of computer plant classification are more in-depth research Correctness of the Identification and classification are significantly increased by using the classifier based on support vector machine with the help of the Eigen values.