Abstract: Existing exemplar-based concealment algorithms have high computational complexity. In order to solve the problem, a real-time spatial error concealment algorithm using segmentation is proposed. The error block is first segmented by its texture characteristic, and each segmented sub-block is recovered by searching the best matching block in texture region. The paper uses this method to realize the concealment. Experimental results show that, the proposed algorithm can provide better image concealment quality while decreasing the computational complexity compared with other classic exemplar-based error concealment algorithm.

Key words: spatial error concealment, image segmentation, exemplar-based error concealment, real-time error concealment, image inpainting

摘要： 基于模板的空间错误隐藏算法计算复杂度较高。为此，提出一种基于图像分割的实时空间错误隐藏算法。将分割后错误块边界的同类纹理作为参考模板，在对应的纹理区域中寻找最佳匹配块，以实现错误隐藏。实验结果表明，与传统算法相比，该算法能获得较好的隐藏效果，且计算量较低。

关键词: 空间错误隐藏, 图像分割, 基于模板的错误隐藏, 实时错误隐藏, 图像修复

CLC Number: 

• TP391