Quantization Schemes for Low Bitrate Compressed Histogram of Gradients Descriptors
Abstract
We study different quantization schemes for the Compressed Histogram of Gradients (CHoG) image feature descriptor. We propose a scheme for compressing distributions called Type Coding, which offers lower complexity and higher compression efficiency compared to tree-based quantization schemes proposed in prior work. We construct optimal Entropy Constrained Vector Quantization (ECVQ) code-books and show that Type Coding comes close to achieving optimal performance. The proposed descriptors are 16× smaller than SIFT and perform on par. We implement the descriptor in a mobile image retrieval system and for a database of 1 million CD, DVD and book covers, we achieve 96% retrieval accuracy using only 4 kilobytes of data per query image.
Cite
Text
Chandrasekhar et al. "Quantization Schemes for Low Bitrate Compressed Histogram of Gradients Descriptors." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2010. doi:10.1109/CVPRW.2010.5543242Markdown
[Chandrasekhar et al. "Quantization Schemes for Low Bitrate Compressed Histogram of Gradients Descriptors." IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops, 2010.](https://mlanthology.org/cvprw/2010/chandrasekhar2010cvprw-quantization/) doi:10.1109/CVPRW.2010.5543242BibTeX
@inproceedings{chandrasekhar2010cvprw-quantization,
title = {{Quantization Schemes for Low Bitrate Compressed Histogram of Gradients Descriptors}},
author = {Chandrasekhar, Vijay and Reznik, Yuriy A. and Takacs, Gabriel and Chen, David M. and Tsai, Sam S. and Grzeszczuk, Radek and Girod, Bernd},
booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops},
year = {2010},
pages = {33-40},
doi = {10.1109/CVPRW.2010.5543242},
url = {https://mlanthology.org/cvprw/2010/chandrasekhar2010cvprw-quantization/}
}