融合方向测度和灰度共生矩阵的纹理特征提取算法研究

为了解决提取图像纹理特征时所遇到的纹理方向抑制问题,提出一种融合方向测度和灰度共生矩阵的纹理特征提取算法。该算法通过灰度共生矩阵,提取图像的Haralick特征,其中包括对比度、相关性、能量、逆差矩等,然后利用方向测度引入权值因子,并将其与所提取的Haralick特征相融合,最后对融合后的各个分量进行高斯归一化处理,获取最终的纹理特征集。实验结果表明,与采用灰度共生矩阵方法相比,该算法可以有效的避免图像纹理方向的抑制,所提取的纹理特征具有更强的图像识别能力,对Brodatz标准纹理库分类的正确率也有一定的提高。

Textural Feature Extraction Algorithm Fused with Direction Measure and GLCM method

To solve the texture direction suppression problem occurring on texture feature extraction of image, this paper proposes a new feature extraction algorithm fused with direction measure and GLCM. The algorithm extracts Haralick textural features based on GLCM, which include contract, correlation, angular second moment and homogeneity. Moreover, weight factors are introduced based on the direction measure. Then, the ultimate textural feature which is normalized by gauss method is extracted by fusing Haralick textural features and weight factors together. The experimental results show that the proposed algorithm can avoid the suppression problem of texture direction effectively and also can easily identify the textural image. The accuracy of classification for Brodatz has a certain improved compared with the GLCM method.