基于粗糙集理论的自适应图像内插

利用粗糙集中的不可分辨理论提出了一种基于粗糙集理论和旋转掩模(rotating mask)相结合的自适应图像内插算法.首先利用图像的连续性,提出了用粗糙集中的上近似集(upper approximation)和下近似集(lower approximation)来确定与缺损点具有不可分辨关系的样本点,然后利用这些同类样本点对缺损点进行自适应的内插.实验结果表明,利用该算法内插的图像质量无论客观评价标准还是主观评价标准均好于传统的图像内插法,该算法是粗糙集理论与图像信息处理相结合的有益尝试.

Adaptive Image Interpolation Based on Rough Set

A new adaptive interpolation algorithm for still image is presented.The algorithm is based on the concept of indiscernibility relation in rough sets theory and incorporates a novel rotating mask.The sample points that are indiscernible to the loss point are first identified by applying the concept of upper and lower approximation based on the continuity of images.Then the lost point is further identified using the information of the sample points.Simulation results demonstrate that the new algorithm improves substantially the subjective and objective quality of the interpolated images over conventional interpolations.The algorithm represents an attempt to incorporate rough sets in image processing.