目的 为进一步提高分割精度，在模糊聚类的基础上引入统计信息，提出一种鲁棒型空间约束的模糊聚类分割算法。方法 基于局部空间信息的先验概率与后验概率，提出一种新型空间约束项，并通过卷积操作提高运行效率；进而引入负对数联合概率作为测度函数，进一步提高算法对于各像素点所属类别的甄别能力；同时将测度函数与空间约束项整合至目标函数中，通过迭代更新各参数达到最小化目标函数的目的。结果 对于合成图像的实验结果表明，本文算法对于噪声类型和噪声强度具有较强的鲁棒性；对于彩色图像的实验结果表明，在适当的特征描述符的辅助下，本文算法也能够获得令人满意的分割结果和较高的分割精度。结论 本文算法克服了现有算法的缺陷，进一步提升了图像的分割精度。其适用于分割带噪声图像，且在适当纹理特征的辅助下分割彩色图像，与同类算法的比较实验结果验证了本文算法的有效性。

Objective Fuzzy clustering-based methods and statistical models have been widely used for image segmentation. To improve segmentation accuracy, this study develops a novel robust spatial factor-based fuzzy clustering algorithm by introducing statistical information into the fuzzy objective function. Method A novel spatial factor is proposed to overcome the impact of noise on images. The proposed spatial factor is constructed based on the posterior and prior probabilities by incorporating the spatial information between neighboring pixels. It acts as a linear filter that smoothens and restores noise-corrupted images. The proposed spatial factor is fast, easy to implement, and capable of preserving details. The negative logarithm joint probability, which serves as a dissimilarity function, considers the prior probabilities and thus improves the capability to identify the class of each pixel. Integrating the dissimilarity function and novel spatial factor into the fuzzy objective function, we can obtain the final segmentations by iteratively minimizing the objective function. Result The comparison results on synthetic images demonstrate that the proposed algorithm can realize accurate segmentation and strong de-noising. The comparison results on color images demonstrate that the proposed algorithm can produce satisfactory segmentation results and accuracies by utilizing the suitable feature descriptor. Conclusion The proposed algorithm address the drawbacks of current segmentation algorithms and further improve the accuracy for image segmentation. It outperforms state-of-the-art segmentation approaches in terms of accuracy. The proposed algorithm applies to image with noise and color image in aid of texture features.