基于Gabor特征和局部二值模式融合的纹理图像识别

传统的Gabor滤波方法和局部二值模式(Local Binary Pattern,LBP)算法提取的纹理特征鉴别能力不足,导致纹理识别精度不够理想.为了解决上述问题,提出一种将全局Gabor特征和局部LBP特征进行融合的纹理图像识别方法.该方法利用Gabor滤波方法提取纹理图像的全局特征,利用LBP算法提取纹理图像的局部特征,然后在最近子空间分类器的框架下实现全局和局部特征的融合以及纹理图像识别.在CURe T和KTH-TIPS基准纹理库上的实验结果表明,笔者提出的方法显著超越了传统Gabor滤波方法和LBP算法的纹理识别精度.

Texture Image Recognition Based on the Fusion of Gabor Features and Local Binary Patterns

Traditional Gabor filtering and local binary pattern( LBP) methods can't get satisfying texture recognition results due to the lack of good discriminative ability. To solve the above problem,a method based on the fusion of global Gabor feature and local LBP feature is proposed to implement the texture recognition. In this method,the global texture feature is extracted by Gabor filtering method,and the local texture feature is extracted by LBP algorithm. Finally,the feature fusion and texture recognition are implemented in the framework of nearest subspace classifier. The experimental results on the benchmark datasets of CUReT and KTH-TIPS show that the proposed method significantly outperforms the traditional Gabor filtering and LBP methods.