基于峰值聚类趋势检验的含噪声图像快速阈值分割方法

Fast Thresholding Segmentation Method Based on Peak Clustering Tendency Test for Noisy Image

为了改善含噪声图像的分割效果并提高分割速度,本文提出了一种基于峰值聚类趋势检验的快速阈值分割方法。首先,根据本文提出的平行线投影分割算法,将二维直方图投影为一维直方图;然后利用基于直方图的加权模糊c均值聚类算法对图像中的像素点进行聚类,并对聚类结果作峰值聚类趋势检验,根据检验结果调整平行线平移量参数,直到找到符合双峰分布特性的直方图;最后根据聚类算法得到的阈值实现图像的分割。理论分析和实验仿真结果表明,该方法能够达到实时性和精确性的要求。

In order to improve segmentation effect of noisy image and segmentation speed, a fast image thresholding segmentation method is presented in this paper. The proposed method is based on peak clustering tendency Test. Firstly ,Parallel projection segmentation algorithm is used to project two-dimension histogram into one-dimension histogram according to initial parallel moving parameter. Then histogram-based weighting Fuzzy c-Means algorithm is used to cluster pixels in image, parallel moving parameter is adjusted according to the result of peak clustering tendency Test, till proper histogram which satisfies double-peak distribution is found. Finally, threshold gotten by clustering algorithm is used to implement fast segmentation. The experimental results show that, the proposed method satisfies real-time performance and accuracy requirements of noisy image segmentation.