基于FMM的图像建模与语义对象分割

Image modeling and semantic object segmentation based on finite mixture model

提出一种基于有限混合模型(FMM)的图像自动建模与语义分割方法。算法采用分水岭算法进行预分割,以给FMM模型设定合适的初始参数。同时针对传统EM算法的不足,对其分类结果自动执行合并与分裂操作以获得最佳分类数并跳出局部极值点。实验结果显示新方法能获得较精确的具有良好视觉感知的语义分割结果。

A novel scheme for automatic image modeling and semantic image segmentation based on finite mixture model （FMM） is proposed. The classical watershed algorithm is firstly employed to implement pre-segmentation and its results are used to initialize model parameters. Then, to overcome the drawbacks of traditional EM algorithm, its classified results are further implemented automatic merging and splitting to achieve optimal number of categories and escape local extrema. Experiments are conducted and the results reveal that the proposed scheme can achieve accurate and perceptually consistent segmentation results.