【摘要】 先验置信传播(priority-BP)算法很难在实际中达到实时处理的要求,计算效率也有很大的提升空间。针对先验BP算法在图像修复上的应用,改进算法主要在信息传递方面,提出改进措施。即在算法迭代过程中快速更新目标区域图像信息的同时,加入对各节点相似程度的比较,对相似程度高的节点完成聚类合并,使节点在后续迭代过程中,节点所能代表的范围增加。相应地节点在组建更新标签集时,所用标签的尺寸也会增加。取得提高标签裁剪效率,加快信息传递收敛速度的目的。将改进算法用于实例验证,其结果表明该算法在不损失图像修复精度同时,提高了图像修复的效率。更多还原

【Abstract】 Absrtact: In practice, the Priority Back Propagation(Priority-BP) algorithm is difficult to meet the requirements of real-time processing, and its computational efficiency also has a lot of room for improvement. In order to apply the priority-BP algorithm in image repairing, in this paper, an improved priority-BP algorithm for image repairing is proposed. In this improved algorithm, the image information of the target area is quickly updated during the iterative process of the algorithm, and at the meantime the comparison of the similarity degree of each node is added, then those nodes with high similarity are clustered and merged, so that the node can increase the range of representation in the subsequent iteration, and the corresponding nodes will increase the size of the label used when create and update label sets, in order to improve the efficiency of label cutting and speed up the convergence of information transmission purposes. The results of example verification show that the improved algorithm can improve the efficiency of image repairing without losing the accuracy of image repairing.更多还原