-
摘要: 针对传统串行特征融合方法中矢量空间维数的限制以及并行复矢量特征融合方法中能够融合的特征类别数量有限的弱点, 提出一种建立在四元数空间中的新型特征并行融合方法. 本文从理论上详细证明了该方法的合理性及其实际应用中的可行性, 并将实数中的 Fisher 鉴别分析法推广到四元数空间, 同时证明了推广 Fisher 鉴别分析法用于图像模式分类的可行性, 并给出使用的具体方法和步骤. 最后将本文提出的推广方法用于人脸检测, 取得了良好效果.Abstract: In view of the dimension limit of traditional serial feature fusion method and the quantity limit of parallel complex vector feature fusion method, an evolution of parallel vector feature fusion method based on quaternion is proposed. The feasibility and rationality of the method are proved in detail. Meanwhile, Fisher classification method in real and complex domains is generalized into quaternion space. Furthermore, the feasibility and the detailed proving process of the generalized Fisher method used in image classification are also given. Finally, the novel algorithm is used in face detection and good effect is obtained.
-
Key words:
- Feature fusion /
- quaternion /
- self-conjugate quaternion matrix /
- classification /
- face detection
计量
- 文章访问数: 3418
- HTML全文浏览量: 119
- PDF下载量: 1291
- 被引次数: 0