摘要
针对纹理检索中的图像特征提取问题,基于小波方向波变换和灰度共生矩阵,提出了一种新的纹理图像特征提取方法,可对纹理图像进行检索.首先通过计算小波方向波变换分解后获得的各子带的均值和标准方差以及灰度共生矩阵的二阶矩、对比度、相关系数、熵的均值构造纹理图像的特征向量,然后采用不同权值的平均欧氏相似性度量方法作为相似度衡量标准进行检索.研究结果表明该方法具有更好的检索效果,平均查准率有较大的提高.
For the problem of image feature extraction in texture retrieval,a method of texture image feature extraction is proposed based on Wavelet-directionlet transform and gray level concurrence matrix(GLCM)and applied to the texture image retrieval.Feature vector of texture image is constructed by calculating the mean and standard deviation of sub-bands after decomposing image by Wavelet-directionlet transform,and the mean of the angular second moment,contrast,correlation,entropy of the GLCM.Then the similarity measurement of average Euclidean with different weights is used for image retrieval.The experiments show that the method has a better retrieval accuracy,and average precision is greatly improved.
出处
《徐州工程学院学报（自然科学版）》
CAS
2016年第4期65-69,共5页
Journal of Xuzhou Institute of Technology(Natural Sciences Edition)
基金
国家自然科学基金项目(61201323)
关键词
图像检索
小波方向波变换
灰度共生矩阵
特征提取
image retrieval
wavelet-directionlet transform
gray level concurrence matrix(GLCM)
feature extraction
