基于欧拉向量的彩色图像检索方法

Color Image Retrieval Methods Based on Euler Vector

为了有效地管理图像数据库,并对其进行查询检索所需的图像特征,提出了一种基于欧拉向量的色彩图像检索方法.该方法将图像位平面的灰度码表示和增量位平面表示相结合用于图像检索.结合方式有两种:一种方式是获得联合欧拉向量作为图像的特征,并对联合欧拉向量进行高斯归一化,然后采用距离公式对图像进行相似度计算,并对彩色图像进行检索;另一种方式是利用图像重要位平面所获得的灰度码位平面和增量位平面,分别获得各自的欧拉向量,采用距离公式加权的方法进行图像相似度计算.实验结果表明该方法有效改善了检索效果.

A compact representation of image features is highly desirable for efficient management of the image database, search, and retrieval. A method of image retrieval using Euler vector is proposed. The method combining gray-coded bit-planes with augmented bit-planes for image retrieval is divided into two ways. The first is to get the united Euler vector as image features, and make use of distance measures to the retrieval of similar color images to query images after Gaussian normalization. The other uses the gray-coded bit-planes based Euler vector and the augmented bit-planes based stacked Euler vector as the image features to compute two corresponding distances between the query image and the object ones, and then assigns a weight to each of the two distances to obtain a new weighted distance with which to retrieve the similar images. Experimental results are reported to demonstrate that the retrieval performance is improved significantly.