LOAD: Local orientation adaptive descriptor for texture and material classification
详细信息 查看全文
- 作者：Xianbiao Qia ; b ; qixianbiao@gmail.com" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Guoying Zhaoa ; gyzhao@ee.oulu.fi" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Linlin Shenb ; llshen@szu.edu.cn" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Qingquan Lib ; liqq@szu.edu.cn" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae ; Matti Pietikä ; inena ; mkp@ee.oulu.fi" class="auth_mail" title="E-mail the corresponding authorAuthor Vitae
- 关键词：Local orientation adaptive descriptor ; Texture classification ; Material recognition ; Improved Fisher vector ; Convolutional neural network
- 刊名：Neurocomputing
- 出版年：2016
- 出版时间：5 April 2016
- 年：2016
- 卷：184
- 期：Complete
- 页码：28-35
- 全文大小：1984 K
文摘
- •
This paper proposes a novel and discriminative texture descriptor (Local Orientation Adaptive Descriptor).
- •
We build a new real-world material data set that contains 13 categories.
- •
We demonstrate the effectiveness of our feature on texture and real-world material classification.
- •
We experimentally demonstrate that the proposed LOAD shows strong complementary property with the learning based features, such as CNN.