基于多特征融合的图像检索方法

提出了一种综合颜色直方图和彩色共生矩阵相结合的彩色图像检索方法.首先,对彩色图像进行量化,并利用直方图计算图像之间的欧式距离;然后,利用彩色共生矩阵提取图像的纹理特征并计算图像之间的欧式距离;最后,利用综合加权的颜色特征和纹理特征实现图像检索.实验结果表明,与直方图和灰度共生矩阵相比,该方法能较好地满足用户需求,具有较高的检索性能.

Image Retrieval Algorithm Based on Multi-feature Fusion

This paper puts forward an image retrieval method by combining comprehensive color histogram and color co-occurrence matrix. Firstly, color images are quantified and the Euclidean distance between the images is calculated by using statistical histogram. Then, by means of color co-occurrence matrix, texture features of image are extracted; and then the Euclidean distance between images can be further calculated. Fi- nally, image retrieval can be accomplished by using the integrated weighted color features and texture fea- tures. The experimental results show that the method proposed has better retrieval performance and can bet- ter meet the needs of users than histogram and gray level co-occurrence matrix.