CAJ | 학술논문

针对传统聚类算法在图像分割中对聚类中心选择敏感，可靠性差的缺点，本文采用 AP 聚类算法研究图像分割问题。AP(Affinity propagation)聚类算法是通过数据点之间的信息传递产生高质量的聚类中心，避免了聚类初始中心选择难的问题。本文通过与 K 均值算法和模糊 C 均值算法在图像分割中的实验比较，得出本算法优于其他两种算法，对图像可取得良好的分割效果。
침대전통취류산법재도상분할중대취류중심선택민감，가고성차적결점，본문채용 AP 취류산법연구도상분할문제。AP(Affinity propagation)취류산법시통과수거점지간적신식전체산생고질량적취류중심，피면료취류초시중심선택난적문제。본문통과여 K 균치산법화모호 C 균치산법재도상분할중적실험비교，득출본산법우우기타량충산법，대도상가취득량호적분할효과。
Aiming at the problem that many conventional clustering algorithms have low reliability due to the fact that-they are very sensitive to the value which is selected for the initial clustering center,the affinity propagation clustering algo-rithm which avoids the difficulty of selecting initial cluster centers by generating high quality centers through the transmis-sion of messages between different data-points is used in this paper for image segmentation.By comparing with experimental results of image segmentation using K-means algorithm and fuzzy C-means algorithm individually,the present algorithm is superior to them and can get much better segmentation results.