一种有效的结构纹理区分方法

提出了一种检索纹理图像的算法.由于纹理图像具有的多样性,处理不同的纹理图像应当采用不同的处理方法.比如,针对结构纹理图像使用小波的方法更为适合,而针对具有随机纹理的图像使用马尔科夫模型效果就更好.因此,在处理图像之前对其按照纹理的情况进行分类合并,就可以有针对性的处理各种纹理图像.我们提出的区分纹理图像的方法基于二值傅里叶谱,首先将纹理库中的图像按结构纹理和随机纹理划分,然后采用多方向小波contourlet提取图像特征来检索图像.实验表明,这种区分算法,可实质性地提高检索率.

An Efficient Method for Discriminating Structural Texture

An image retrieval algorithm for texture classification is presented here.Due to the diversity of the textures,it is more reasonable to apply different methods to process different types of texture images.For example,wavelets are more suitable to structural texture images,while Markov model method does a better job for random texture images.Here a method based on binary Fourier spectrum is proposed to identify structural texture images,whose texture property is more regular.Then the multi-directional wavelet contourlet is applied to extract the features of the images for image retrieval.Simulation results clearly show that the method can improve the image retrieval rate effectively.