大规模多视角高维图像特征提取

项目名称： 大规模多视角高维图像特征提取

项目编号： No.61773328

项目类型： 面上项目

立项/批准年度： 2018

项目学科： 自动化技术、计算机技术

项目作者： 黄伟强

作者单位： 香港理工大学深圳研究院

项目金额： 16万元

中文摘要： 大规模多视角高维图像已经成为大数据的重要组成部分。现有的多视角学习方法仅能用于小型数据集而不适用于大规模多视角高维图像，故如何高效地提取大规模多视角图像特征用于识别与检索是当前亟待解决的问题。为获得有效的特征并适应大数据的存储与快速搜索需求，我们提出双线性投影哈希的思想来发展多视角离散哈希特征提取理论与算法框架。本项目的研究将进一步丰富大规模多视角图像特征提取理论，提高特征提取的有效性并节省计算量与存储空间以适应大规模图像数据处理需求。本项目的研究成果不仅可用于大规模多视角高维图像分析领域，也可拓展到大规模多模态数据、大数据图像特征提取与搜索等领域，具有重要的理论意义和广泛的应用前景。

中文关键词： 线性判别分析；子空间学习；特征提取；大规模；多视角图像

英文摘要： Large scale multi-view high dimensional images have become an important part of big data. As many existing multi-view learning methods can only be used for small data sets and thus not suitable for large-scale multi-view high dimensional image, how to efficiently extract the large-scale multi view image features for recognition and retrieval is the current problems to be solved. In order to obtain the effective features and meet the needs of large data storage and fast search, we propose the idea of bilinear projection hash to develop multi-view discrete hash feature extraction theory and algorithm framework. The research of this project will further enrich the theory of large-scale multi view image feature extraction, and improve the effectiveness of feature extraction, and save the computation and storage space to meet the needs of large-scale image data processing. The research results of this project not only can be used for large-scale multi-view high dimensional image analysis, but also extended to the field of large-scale multi modal data, image feature extraction and data search, and thus generate great theoretical significance and wide application prospect.

英文关键词： linear discriminant analysis;subspace learning;feature extraction;large scale;multi-view image