基于颜色和形状特征的图像检索系统

Image Retrieval System Based on Color and Shape Features

随着信息技术的迅速发展,为了高效地从大量图像实例中找出有用信息,改善基于单一特征的图像检索方法的效率和准确性,提出基于颜色和形状信息的多特征图像检索方法,利用VC++6.0工具开发图像检索系统。算法采用RGB颜色空间分块直方图和边缘形状Sobel算子相结合的算法提取图像特征。利用600幅756×504大小的JPEG四种分类场景图片构建图像库,通过实验,首先基于单一的颜色形状特征进行检索得到结果;再通过综合颜色和形状特征的检索算法实现基于多特征的检索,两次实验相对比结果表明多特征检索系统在实际应用中具有高效性和优越性。

Along with the development of information technology, because the retrieval method based on single feature is lack of efficien- cy and accuracy, for the purpose of finding valuable information from lots of images, a new image retrieval method by comprehensive features is created and using VC＋＋6.0 to develop image retrieval system. The algorithm uses RGB color space block histogram and edge shape detection of Sobel operator to extract image features. The lib is made of 756 × 504 size JPEG images which include four classifica- tion scene images. First based on single feature and then through comprehensive retrieval algorithm based on the characteristics of image retrieval,obtain results of the two experiments, from which show the system＇s efficiency and superiority.