摘要
针对图像检索的多特征融合问题,提出了一种综合BTC颜色矩和灰度共生矩阵的图像检索算法。该算法借鉴BTC的思想,对传统的颜色矩进行改进,将图像分为若干子块,每个子块按照阈值分割为两类像素,计算颜色矩,提取图像的颜色特征;利用灰度共生矩阵提取图像的纹理特征;通过颜色、纹理单特征相似度加权求和来表示图像的综合特征相似度,进行图像检索。实验结果表明,该算法能够综合颜色特征和纹理特征,有效地提高图像检索的查准率。
A new image retrieval method that integrates the BTC color moment and GLCM is proposed for feature fusion. Firstly, using BTC moment to extract color feature, the method improves the traditional color moment. Every image is divided into several image blocks and every block pixels are di- vided into two sets. Then compute the color moment. Secondly, the texture feature is extracted by GLCM. Finally, it can retrieve image using the sum of weighted color feature and weighted texture feature. The results of experiments show that combined features can integrate color feature and texture feature. It can also gain better retrieval results among all kind of images.
出处
《电视技术》
北大核心
2012年第11期30-33,40,共5页
Video Engineering
基金
江苏高校优势学科建设工程资助项目
