基于各向异性扩散的彩色细胞图像自动分割

本文采用基于各向异性扩散与均值位移相结合的分割算法对医学图像分割进行研究。将图像分成色度和非色度信道,分别对这两个信道进行各向异性扩散,把扩散后生成的结果进行平均值位移聚类分割,得到图像的最终分割结果。这一方法在有效抑制过分割现象的同时,保留了面积很小但对比度高的感兴趣区。实验表明该算法简单有效且稳定,并以癌细胞为例,给出分割结果。

Automatic Segmentation of Color Cells Based on Anisotropic Diffusion

Medival images segmentation is studied by combining anisotropic diffusion and mean shift technique. Images are splitted into chromatic and achromatic channels and separately smoothed through anisotropic diffusion. The results of the two diffusions are separately segmented by mean shift techniques and their combination yields the final image partition into homogeneous regions. Some experimental results applied to cancer cell images are reported in the paper, which verifies the effectiveness of the proposed technique.