结合形状先验的图割目标分割方法

为了减少图像目标分割过程中噪声、阴影、背景复杂等因素的影响，将形状先验引入图割图像分割框架中，提出一种结合形状先验的图割目标分割方法。该方法给出了形状先验的定义，并将其转化为势函数的形式加入到能量函数中，通过能量最小化过程得到最终分割结果。同时，图像对齐过程能够适应形状模版与待分割目标之间的仿射不同。形状先验能够较好地约束目标边界，相比与传统图割算法，分割结果有了明显改善。实验结果表明，该算法具有有效性。

Graph cuts based object segmentation method combining shape prior

In order to reduce the influences of noises, shadows, complex background in object segmentation, shape prior is introduced into the graph cuts framework, and a graph cuts based object segmentation method combining shape prior is proposed. In this method, the definition of shape prior is provided, which is converted to potential function and added to the energy function, and the final segmentation is obtained by energy minimization. At the same time, the image alignment process can adept to affine variations between the shape template and the object to be segmented. Shape prior can constrain the object boundary well, comparing the traditional graph cuts method, the segmentation results are obviously improved. The experiment results show that the proposed method is effective.