Abstract: An image mosaics algorithm based on salient region matching is proposed by combining the visual attention model and the Mean Value Seamless Cloning（MVSC） algorithm. The new algorithm uses the improved visual attention model to extract the salient region, and the use of matching technology achieves the image matching, and MVSC algorithm has good image fusion character, so that the image mosaics are studied. Experimental results show that this paper algorithm can not only automatically and accurately extract the salient regions, but also reduce the complexity of the image matching, improve the quality of image mosaics.

Key words: salient region, image matching, Mean Value Seamless Cloning（MVSC）, panoramic image

摘要： 将视觉关注模型和均值无缝克隆（MVSC）算法相结合，提出了一种采用显著区域匹配的图像拼接算法。新算法通过改进视觉关注模型提取显著区域，同时利用区域匹配技术实现重叠显著区域的匹配，结合MVSC具有良好的图像融合特性，对图像拼接进行了研究。实验结果表明，该算法不仅可以自动准确地提取显著区域，而且提高了图像匹配的精确度，改善了图像拼接的质量。

关键词: 显著区域, 图像匹配, 均值无缝克隆（MVSC）, 全景图像