Research On Segmentation Of Plant Image Based On NCut

Plant leaves and petals organization, to study the growth and development of plants, andplant health is of great importance. Often use image acquisition technology can be veryconvenient and can be more intuitive for images of the whole plant information or imageinformation for part of the plant, but only for the image is not enough, also need to plant theinformation in the image, image segmentation effectively and accurately for plant leaves, thepetals of the image information, to exclude the interference of irrelevant information, and thenfor the analysis of plant growth and health, which has laid a good foundation in the leaves,petals information, image segmentation method is the most effective way.So far, there have been many algorithms to realize the image segmentation, and differentalgorithms can be showed in some respects and other advantages of the algorithm.Normalized image segmentation (Normalized cut, NCut) algorithm, is by using a weightedundirected graph representation method of image, each pixel in the image as a weightedundirected graph of a point, and by means of the criterion of segmentation based on graphtheory to image segmentation. NCut method compared with other image segmentationmethod, has obvious advantages, make full use of the grey value of image, the gradientinformation, geometric shape, texture information such as lower order and higher orderinformation, and these information effectively together, complement each other. So NCutalgorithms to get the favour of many research institutes, by various researchers’ attention.This article will NCut method applied in the field of plant image segmentation, andcomparing with other methods, analysis of the advantages of the algorithm. This article mainresearch content is as follows:1. The first studied the plant image preprocessing, mainly according to the characteristicsof the plant leaves and petals, respectively, using three methods for the processing of theimage, the first method is gray, the second method is to image filter processing, and the thirdway is by image enhancement processing, according to the related studies three methodsrespectively, and using the plant image related experiments, different methods for comparison.2. The use of level set method based on area and single plant leaf petals segmentationresearch. Analysis of several traditional segmentation method first, and then puts forward themethod based on active contour model, and specific to the level set method, algorithm modelof the level set method is studied, and the leaves and petals image segmentation experiments, tested parameters and the robustness of the algorithm, and algorithm of comparison, algorithmanalysis of existing problems, and puts forward the characteristics of using this method indifferent color space segmentation.3. The paper introduces the process and principle of NCut algorithms, etc., and then onthe leaves, petals image algorithm validation, the results showed that the NCut can have morepetals and leaves of the flower petals in acquiring, or a leaf, and contrast experiment wascarried out in front of the segmentation methods.