一种纹理特征提取方法在图象检索中的应用

Application of textural property extraction method in image retrieval

针对利用Gabor小波变换提取图象纹理特征检索效率和检索速度比较低的缺点,提出了基于Gabor滤波器和Gabor小波变换提取纹理特征的分析方法之上的Gabor小波归一化的思想,使各特征分量具有相同的权重,减少了计算量,有效的提高了图象的检索速度和准确度,提供了在没有人参与的情况下机器能自动识别或理解图象重要特征的可能,并设计了一个原形系统,以验证理论的准确性。

In view of the that retrieval efficiency and speed are very low using the Gabor wavelet transformation to extract image textural property, this paper proposes the Gabor wavelet normalization view based on the analysis of the methods of using the Gabor filter and the Gabor wavelet transformation extracting textural property. This view enables various characteristic components to have the same weight, reduce the computation load and raise speed and accuracy of retrieving image.It also provides the possibility that the machine could identify or understand image key characters automatically when nobody works, and design a primary system to confirm accuracy of this theory.