欢迎您！东篱公司
退出
-
申报数据库
-
立项数据库
-
成果数据库
-
项目获奖数据库
中文摘要：
通过提取熵和标准差两个特征对图像进行分类，将分类结果采用PNN算法进行矢量量化，将其生成的码书作为LBG算法的初始码书以降低该算法对初始码书敏感的缺点。同时在水印提取过程中，首先对码书进行预处理以减少图像攻击对码书的影响，再提取水印图像。实验证明，使用该方法不仅得到了较高的图像质量，而且对常见的攻击也具有鲁棒性。另外，在传输过程中对码书和索引值的窜改，也具有一定的抗攻击性。
英文摘要：
Image region is classified by the entropy and the standard deviation, and then quantized by PNN algorithm separately. Additionally, used the eodebook generated by PNN algorithm to the initial codebook of LBG algorithm for reducing LBG algorithm＇ s sensitivity. Particularly, in watermark extracting process, preprocessed the codebook to decrease influence on eodebook. Experimental results show that the proposed algorithm not only improves codebook quality evidently, but also is robust to common image processing operations. In addition, the proposed algorithm has some anti-attack capability for tampering with codebook and index values.
同期刊论文项目
同项目期刊论文
期刊信息
位置：