本文選題：圖像分割 + 模糊聚類�。� 參考：《控制與決策》2017年02期

【摘要】：為解決傳統(tǒng)FLICM算法需人為給定圖像聚類數(shù)的問題,基于該算法通過聚類中心描述聚類的特點,設(shè)計以聚類中心為操作對象的分裂合并操作,以實現(xiàn)可變類圖像分割.在此基礎(chǔ)上定義分裂合并操作的接受率,不但能夠有效避免算法陷入局部極值,促進(jìn)其快速收斂,同時有利于參數(shù)閾值的自適應(yīng).分別利用所提出算法和傳統(tǒng)ISODATA算法分割模擬圖像和灰度紋理圖像,對其結(jié)果的定性定量分析驗證了所提出算法的有效性和普適性.
[Abstract]:In order to solve the problem that the traditional FLICM algorithm needs a given number of images, based on the characteristics of the clustering described by the clustering center, the splitting and merging operation based on the clustering center is designed to realize the variable image segmentation. On this basis, the acceptance rate of splitting and merging operations is defined, which can not only effectively avoid the algorithm falling into local extremum, promote its fast convergence, but also facilitate the self-adaptation of parameter threshold. The proposed algorithm and the traditional ISODATA algorithm are used to segment the simulated image and the gray texture image respectively. The results are qualitatively and quantitatively analyzed to verify the effectiveness and universality of the proposed algorithm.
【作者單位】： 遼寧工程技術(shù)大學(xué)測繪與地理科學(xué)學(xué)院;
【基金】：國家自然科學(xué)基金項目(41271435,41301479) 遼寧省自然科學(xué)基金項目(2015020190)
【分類號】：TP391.41
，

本文編號：2062984