摘要
为解决图像检索领域现有的特征提取方法信息保留不足,检索方法单一,以及时效性不强等问题,对一些独特的图像分析方法进行讨论。方法:利用一种新的图像特征侦测方法生成图像特征库,然后利用该方法提取检索图像内容特征,经过相似性度量准则度量后,最终以度量排序结果进行图像检索。结果:实验结果表明,与多种流行的图像检索方法相比,无论是在查准率还是在查全率方面该方法都具备较好的性能。结论:所述图像检索方法利用一种新的各向异性的微结构分析方法实现了对图像特征的描述,有效提高了图像检索能力,具有很强的实用性。
The objective of this paper is to discuss a new approach to image retrieval with the progress in improving the time effectiveness and the ability of the extracted features in representing more information hidden in images. We use an anisotropy method based on hybrid micro-structure to detect features of images in constructing the feature library of all the images, and we compare features of the image to be retrieved with the features in library to obtain output. The experimental result shows that the method based on hybrid micro-structure analysis has a good performance to achieve higher ratio of recall and precision. The conclusion can be drawn that the hybrid micro-structure analysis is a practical and effective method to describe features of images in image retrieval.
出处
《电脑与信息技术》
2016年第4期1-5,共5页
Computer and Information Technology
关键词
图像检索
微结构
特征提取
边缘方向
直方图
image retrieval
micro-structure
feature extraction
edge orientation
histogram
