优选抑制式非局部空间模糊C-均值图像分割方法

当图像被噪声严重污染时,像素的邻域像素也可能被污染。此时,来自于像素点的邻域像素的局部空间信息无法在含噪图像分割中发挥积极的指导作用。鉴于此,利用图像中与像素具有相似邻域结构的像素构造新的非局部加权和图像,并在新图像的灰度直方图上采用优选抑制式模糊C-均值聚类,提出优选抑制式非局部空间模糊C-均值图像分割方法。实验结果表明,该方法能进一步提高模糊C-均值聚类方法对于图像噪声的鲁棒性,获得了更加理想的分割结果。

Selection-suppressed non-local spatial FCM image segmentation method

When the image is heavily contaminated by noise, the adjacent pixels of a pixel may be also corrupted by noise. Under this condition, the local spatial information derived from the adjacent pixels of the giuen pixel cannot play a positive part in guiding noisy image segmentation. In order to solve this problem, this paper proposed a selection-suppressed non-local spatial FCM image segmentation method. This method firstly constructed the non-local weighted-sum image by using a set of pixels with a similar neighborhood configuration of the pixel in the image, and then performed a selection-suppressed FCM algorithm on the histogram of the obtained image. Segmentation experiments show that the proposed method further improves the robustness of FCM method to image noise and obtains more perfect image segmentation results.