基于边缘方向投影的图像块修复方法

压缩图像在传输中,信道干扰常导致图像块数据丢失.给出一种基于边缘方向投影的图像损失块修复方法.按照周围邻域未损失像素的边缘方向信息将损失块分类,并根据类别用结合频域和空域信息的凸集投影算法自适应地修复损失块.能较好地修复图像边缘和复杂纹理.与RIBMAP方法相比,此方法在修复质量和算法顽健性上均有提高.

Recovery of image blocks using the method of edge orientation projections

In compressed image transmission, channel interference often results in data missing in image blocks. A method, was presented based on edge orientation projections for recovery of missing image blocks. The missing block was classified through edge orientation information of the surrounding neighborhood, and was recovered using POCS algorithm by combining frequency and spatial domain information according the class. The proposed method can restore image edge and complex texture satisfactorily. Experimental comparison with RIBMAP shows obvious improvement of the proposed method in restored image quality and robustness.