Loading [a11y]/accessibility-menu.js
Gradient preserving quantization | IEEE Conference Publication | IEEE Xplore

Gradient preserving quantization

Publisher: IEEE

Abstract:

Local features are widely used for content-based image retrieval and object recognition. Most feature descriptors are calculated from the gradients of a canonical patch a...View more

Abstract:

Local features are widely used for content-based image retrieval and object recognition. Most feature descriptors are calculated from the gradients of a canonical patch around repeatable keypoints in the image. In this paper, we propose a technique for designing quantization matrices that reduce the mean squared error distortion of the gradient derived from DCT-encoded canonical patches. Experimental results demonstrate that our proposed patch encoder greatly outperforms a JPEG encoder at the same encoding complexity. Moreover, our quantization matrices achieve lower gradient distortion and larger number of feature matches at the same bit-rate.
Date of Conference: 30 September 2012 - 03 October 2012
Date Added to IEEE Xplore: 21 February 2013
ISBN Information:

ISSN Information:

Publisher: IEEE
Conference Location: Orlando, FL, USA

References

References is not available for this document.