高级检索

基于方向经验模型分解的纹理合成

Texture Synthesis Based on Directional Empirical Mode Decomposition

  • 摘要: 提出的纹理合成算法首先将样本纹理通过方向经验模型分解求得其固有方向;然后在该方向上进行方向经验模型分解,获得一系列分解图像;在各级图像上应用方向经验模型分解提取出的特征值来搜索匹配块,采用相关搜索来完成各级图像的合成;将其进行合并,得到在该方向上的合成图像;最后沿固有方向进行逆旋转变换,得到最终的合成结果.实验结果表明,该算法不仅合成质量高,而且简单、快速.

     

    Abstract: Texture is an essential aspect in virtual reality,which makes texture synthesis an important research.This paper proposes a novel texture synthesis algorithm.In this algorithm,the signature direction of a sample texture is obtained by using DEMD (direction empirical mode decomposition) method.Then the calculated direction is processed into a set of decomposed images by using direction EMD approach.The feature values extracted from all the levels of the decomposed images are used to search matching patterns.The synthesized image is formed by using a correlation searching.Thus a synthesis associated with the direction is made after composing all levels of decomposed images.In the end the final image can be synthesized by rotating the image along the calculated direction.Our experiment results show that this algorithm is simple, fast and delivers high quality synthesis.

     

/

返回文章
返回