Abstract: The lack of considering aesthetic values is the major drawback of existing image resizing algorithms. To deal with this problem, a content aware image resizing algorithm using aesthetic rules is proposed. Firstly, the importance map of the original image is computed via image co-segmentation and visual saliency detection so as to prevent important objects in the image from distortion in the subsequent treatment process. Then, specific aesthetic rules and the corresponding quantization formulas for two common types of images in a mobile terminal are established. Finally, the algorithm performs the operations of seam carving using the aesthetic rules and the importance map to resize the image. The results show that the proposed algorithm, while preserving image information, achieves better performance in terms of aesthetics than a number of other image resizing algorithms.

摘要： 针对目前内容感知的图像缩放算法存在美学效果考虑不足的问题，提出一种结合美学原则的缩放算法。算法首先采用图像协同分割的思想并结合视觉显著性检测获取图像的重要度图，以此作为图像重要内容的依据，在后续处理中防止其变形；其次针对移动终端常见的两类图像分别选择相适应的美学原则并建立其量化公式；借鉴经典的Seam Carving算法思想，利用美学构图原则和重要度图来指导裁剪线的复制与删除，达到图像缩放的效果。实验结果表明，与同类算法相比，该算法的缩放结果在保留原图重要信息的条件下，更具美感。

关键词: 图像缩放, 内容感知, 美学原则, Seam Carving算法