摘要:近年来，随着互联网高速发展，网上的图片信息急剧增加，这些数据信息大都以图像为主。如何有效地组织、管理和检索大规模的图像数据已成为迫切需要解决的问题。于是基于颜色的图像检索作为一个崭新的研究领域呈现在人们面前。
基于颜色的图像检索技术的应用使管理者从大量的、单调的人工管理工作中解放出来，能够方便、快速、准确的从图像数据库中查找特定图像。CBIR技术的核心是表示图像颜色的特征，而颜色特征计算简单，性质稳定，作为图像的一种重要视觉信息，在中已得到广泛应用。本文介绍了一种基于颜色特征的图像检索技术研究方法。
本系统通过将基于改进的加权的局域颜色直方图的图像检索方法和全局直方图的图像检索方法相结合，提高查全率和查准率。其中，基于分块局域直方图的检索方法利用了图像中间部分的重要性，将图像平均划分成若干个子块，取中间一块的图像，计算其与参考位图相应位置的颜色特征距离，再计算原图的颜色直方图与参考位图的颜色特征距离，分别赋予权值后得出的值就是图像之间颜色的相似程度。本文引入欧氏距离的相似性度量方法实现图像检索。实验表明，该方法具有较好的查全率和查准率。
关键词 ：图片检索；网络数据；颜色直方图；几何距离
Abstract:In recent years, with the rapid development of Internet, online picture information increased dramatically, these data are mostly based on the image. How to effectively organize, manage, and retrieve large amounts of image data has become an urgent problem to be solved. Therefore, based on color image retrieval, presenting in front of people a new field of study.
Color image retrieval based on a lot of tedious manual management from freed, enables managers to find a particular image, convenient, fast and accurately from the image database. Wherein said image content, a simple, stable color characteristics, has been widely used as an important visual image information of CBIR core technology inch paper describes a method of color-based image retrieval.
The system is weighted by improving local color histogram-based image retrieval method based on global image retrieval method to improve the combination of recall and precision. Characterized in that the block retrieval method based on local histogram, using the image of the intermediate portion of the importance of the average image is divided into a plurality of sub-blocks, take the middle piece of the corresponding position in the bitmap image color characteristic distance calculated with reference to the original color histogram and color characteristics of the reference bitmap from the calculation of the weights were given the value obtained was the degree of similarity between the images. This article describes the Euclidean distance similarity measure for image retrieval. Experiments show that the method has good recall and precision.
Keywords Image retrieval network data color histogram geometric distance
