摘要
为提高乳腺组织图像特征提取的正确率,结合灰度共生矩阵纹理特征与Tamura纹理特征,给出一种联合提取与识别算法。采用双边滤波、受约束限制自适应直方图均衡和L0梯度滤波,对数字乳腺层析图像进行预处理,滤除其噪声并提高对比度;同时考虑灰度共生矩阵与Tamura纹理特征,通过增加特征值维度,将两者融合入一个矩阵之中,并依据特征选择规则提取最适特征;用支持向量机对特征向量分类。与仅依赖其中一种纹理特征的提取算法相比,所给算法可提高特征提取与识别正确率,从而更好地实现图像分类。
In order to improve the texture feature extraction accuracy of digital breast tomosynthesis image,ajoint extraction and recognition algorithm is proposed by combining gray level co-occurrence matrix texture feature with Tamura texture feature.By using bilateral filtering,contrast-limited adaptive histogram equalization and L0 gradient filtering,the digital breast tomosynthesis image is preprocessed to filter its noise and improve its contrast.The gray level co-occurrence matrix and Tamura texture feature are considered at the same time,and by adding the dimension of eigenvalue,they are integrated into one matrix,then,the optimal feature is selected out according to the feature selection rules.Support vector machine（SVM）is used to classify feature vectors.Compared with the similar extraction algorithm which only depends on one of the texture features,the proposed algorithm can improve the accuracy of feature extraction and recognition and thus achieve better image classification.
作者
汪友明
张菡玫
汤少杰
WANG Youming;ZHANG Hanmei;TANG Shaojie(School of Automation,Xi′an University of Posts and Telecommunications,Xi＇an 710121,China)
出处
《西安邮电大学学报》
2018年第3期65-68,共4页
Journal of Xi’an University of Posts and Telecommunications
