Constrained Mutual Convex Cone Method for Image Set Based Recognition
Constrained Mutual Convex Cone Method for Image Set Based Recognition
复制标题
DOI：
10.1016/j.patcog.2021.108190
复制
发表时间：
2019-03
期刊：
影响因子：
--
通讯作者：
Naoya Sogi;Rui Zhu;Jing-Hao Xue;K. Fukui
中科院分区：
文献类型：
--
作者：
Naoya Sogi;Rui Zhu;Jing-Hao Xue;K. Fukui
在本文中，我们提出了基于凸锥的图像集分类框架。图像集分类旨在将通常从视频帧或多视图摄像机获得的一组图像分类为目标对象。为了准确、稳定地对集合进行分类，必须准确地表示集合的结构信息。图像特征有多种，例如基于直方图的特征和卷积神经网络特征。我们应该注意到，它们中的大多数都具有非负性，因此可以有效地用凸锥体来表示。这导致我们将凸锥表示引入图像集分类。为了建立基于凸锥的框架，我们在数学上定义两个凸锥之间的多个角度，然后使用这些角度来定义它们之间的几何相似性。此外，为了增强框架，我们引入了两个判别空间。我们首先提出一个判别空间，最大化锥体之间的间隙并最小化类内方差。然后，我们通过在间隙上引入权重来将其扩展到加权判别空间以处理复杂的数据分布。此外，为了降低所提出方法的计算成本，我们开发了一种快速实施的新策略。通过使用五个数据库进行实验证明了所提出方法的有效性。
In this paper, we propose convex cone-based frameworks for image-set classification. Image-set classification aims to classify a set of images, usually obtained from video frames or multi-view cameras, into a target object. To accurately and stably classify a set, it is essential to accurately represent structural information of the set. There are various image features, such as histogram-based features and convolutional neural network features. We should note that most of them have non-negativity and thus can be effectively represented by a convex cone. This leads us to introduce the convex cone representation to image-set classification. To establish a convex cone-based framework, we mathematically define multiple angles between two convex cones, and then use the angles to define the geometric similarity between them. Moreover, to enhance the framework, we introduce two discriminant spaces. We first propose a discriminant space that maximizes gaps between cones and minimizes the within-class variance. We then extend it to a weighted discriminant space by introducing weights on the gaps to deal with complicated data distribution. In addition, to reduce the computational cost of the proposed methods, we develop a novel strategy for fast implementation. The effectiveness of the proposed methods is demonstrated experimentally by using five databases.