Segmentation Model for the Random Image of Navel Orange Pest Based on Local Information Implanting and Improved Fuzzy Mean Value Algorithm

摘要

In order to develop the recognition technology for general machine of different navel orange pest, the blue component image segmentation with the background being removed for navel orange image after pests damage is carried out. Firstly, according to the feature that the fuzzy C- means has robustness to the image noise, the spatial local information is incorporated into the objective function to improve the robustness of the algorithm. At the same time, based on intuitionistic fuzzy set theory, the algorithm introduces the concepts of non-membership degree and hesitation degree, which makes the definition of the membership degree matrix more reasonable. Image segmentation experiments based on noise-free image and Gauss noise show that the proposed algorithm is not sensitive to image noise and it is also able to overcome the shortcomings of the traditional algorithm for image segmentation. Additionally, the algorithm is basically consistent with the traditional algorithm in terms of the time complexity, therefore, it can be effectively used for the segmentation of noisy images. The orange image segmentation algorithm of marking the reconstruction gradient for the internal and external constraints is used to extract the pest damage boundary with less missing and false detection, the complete and clear boundary contour and good overall effect. Moreover, it is able to obtain more ideal image segmentation.