Abstract: In order to solve the authenticity identification algorithm’s problem of low accuracy and large amount of calculation in Chinese painting based on pixel level, this paper proposes a regional level belief propagation algorithm. The image is divided into blocks which is introduced into the basic belief propagation algorithm, excluding non-important information in the transmission process of regional block message. Through the region units and eight surrounding directions, confidence function can be iteratively calculated in the basis of information transmitting and the authenticity of the image block can be reasoned, and then the state of the authenticity of all kinds of subjects is determined. The experiment results show that the improved algorithm can promote the original algorithm for the accuracy of the image recognition and the algorithm’s operation rate.

Key words: pixel, authenticity identification, belief propagation algorithm, regional level

摘要： 针对像素的国画真伪鉴别算法准确率低、计算量大的问题，提出了一种区域级置信传播算法。将图像分块的思想引入到基本置信传播算法中，在区域块消息的传递过程中剔除非重要信息，以区域块及其周围八邻域传递的消息为基础迭代计算区域的置信度函数，推理出图像块的真伪状态，进而确定各类实验对象的真伪状态。实验结果表明，改进算法提高了原算法对图像识别的准确率和算法的运算速率。

关键词: 像素, 真伪鉴别, 置信传播算法, 区域级