Presentation 2023-01-11
Improvement of Composite-SVM in Hyperspectral Image Classification
Tamura Akito, Kitamura Takuya,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) In this paper, we propose an improved method of composite support vector machines for hyper-spectral image classification problems. In the proposed method, the pixels centered on an input pixel are filterd by a 3D filter. Then, the filter is optimized by support vector machines using composite kernel. Thus, an input pixel can be classified into a class by considering the surrounding pixels. In addition, by including the generated data by inverting or randomly changing the placement of the surrounding pixels in training, the generalization capability can be high regardless of the filtered image direction.
Keyword(in Japanese) (See Japanese page)
Keyword(in English) Composite kernel / Hyper-spectral Image / Support Vector Machines
Paper # MSS2022-61,SS2022-46
Date of Issue 2023-01-03 (MSS, SS)

Conference Information
Committee MSS / SS
Conference Date 2023/1/10(2days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair Atsuo Ozaki(Osaka Inst. of Tech.) / Kozo Okano(Shinshu Univ.)
Vice Chair Shingo Yamaguchi(Yamaguchi Univ.) / Yoshiki Higo(Osaka Univ.)
Secretary Shingo Yamaguchi(Hokkaido Univ.) / Yoshiki Higo(NEC)
Assistant Masato Shirai(Shimane Univ.) / Shinsuke Matsumoto(Osaka Univ.)

Paper Information
Registration To Technical Committee on Mathematical Systems Science and its Applications / Technical Committee on Software Science
Language JPN
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) Improvement of Composite-SVM in Hyperspectral Image Classification
Sub Title (in English)
Keyword(1) Composite kernel
Keyword(2) Hyper-spectral Image
Keyword(3) Support Vector Machines
1st Author's Name Tamura Akito
1st Author's Affiliation National Institute of Technology, Toyama College(NIT)
2nd Author's Name Kitamura Takuya
2nd Author's Affiliation National Institute of Technology, Toyama College(NIT)
Date 2023-01-11
Paper # MSS2022-61,SS2022-46
Volume (vol) vol.122
Number (no) MSS-329,SS-330
Page pp.pp.96-100(MSS), pp.96-100(SS),
#Pages 5
Date of Issue 2023-01-03 (MSS, SS)