Low Bit-rate 3D feature descriptors for depth data from Kinect-style sensors

详细信息    查看全文

• 作者：Sai Manoj Prakhya^a ; ^{saimanoj001@e.ntu.edu.sg} ; Weisi Lin^a ; ^{wslin@ntu.edu.sg} ; Vijay Chandrasekhar^a ; ^b ; ^{vijay@i2r.a-star.edu.sg} ; Bingbing Liu^b ; ^{bliu@i2r.a-star.edu.sg} ; Jie Lin^b ; ^{lin-j@i2r.a-star.edu.sg}
• 关键词：3D feature descriptors ; Compression ; Lattice quantization ; 3D keypoint matching
• 刊名：Signal Processing: Image Communication
• 出版年：2017
• 出版时间：February 2017
• 年：2017
• 卷：51
• 期：Complete
• 页码：40-49
• 全文大小：959 K
• 卷排序：51

文摘

We employ lattice quantization to effciently compress 3D feature descriptors. They can be directly matched in compressed domain. Hence drastically reducing the memory footprint and computational requirements. We propose double stage lattice quantization to compress SHOT descriptor even more. Three state-of-the-art descriptors are compressed and their performance is reported.