International Journal of Applied Mathematics and Machine LearningAuthor's: Duanliang Cao
Pages: [1] - [13]
Received Date: November 16, 2018
Submitted by: Jianqiang Gao.
DOI: http://dx.doi.org/10.18642/ijamml_7100122024
Fuzzy C-means (FCM) algorithm is a widely used clustering, however, it is influenced by the initial cluster centers, and is easy to fall into local opitma. In order to overcome this drawback, we propose a modified cuckcoo search (MCS) which changes the detection probability P from a constant value to a varible number of iterations decreases. This will not only improve the quality of the population, but also ensure the covergence of the algorithm. Therefore, we can use the MCS algorithm to generate the FCM clustering centers and avoid the local optimal problem of FCM. The proposed algorithm has better clustering effect and faster running speed. In this paper, MCS_FCM algorithm classifies the image pixels into a number of categories according to the similarity principle, making the similar pixels clustering into the same category as possible. According to the provided gray value of the pixels to be inpainted, we calculate the category whose distance is the nearest to the inpainting area and this category is to be inpainting area, and then the inpainting area is restored by the modified CDD model to realize image inpainting.
image inpainting (restoration), modified cuckoo search (MCS), fuzzy C-means (FCM), MCS_FCM, modified CDD model.
