Image Compression and Retrieval using Hilbert Curve
- Paper number
IAC-05-B1.P.09
- Author
Mr. Noritatsu Matsuo, Waseda Universiy, Japan
- Coauthor
Mr. Sei-ichiro Kamata, Waseda Universiy, Japan
- Coauthor
Mr. Kazuyuki Tsuneyoshi, Japan
- Year
2005
- Abstract
Huge amount of high resolution satellite images for remote sensing are available in a couple of decades. Image compression technique is required or highly desired by this kind of application where images are obtained at great cost or will be (often intensively) further processed. Image coding standards such as JPEG and MPEG are widely used for compressing these images. However, the JPEG (or MPEG) compressed data stream can not be processed without decoding. In this paper, we concentrate on image compression and retrieval without decoding using Hilbert curve. Hilbert curve is one of space filling curves which pass through all points in a space. When we scan an image along with a Hilbert curve, we obtain the one-dimensional data having the neighborhood property. Our compression method is based on a simple segmentation of this data using a linear interpolation. The merits of our compression method is that (1) the compression ratio is almost equivalent to JPEG2000, (2) the computation is quite simple, (3) the required memory is rather small. This paper presents that our compression algorithm can be utilized for image retrieval from image database without decoding the compressed data stream. The block color histogram generated from the compressed data stream is used to compute the similarity measure between two images. The experimental results show that our approaches achieve better performance than conventional techniques, such as BTC based method.
- Abstract document
- Manuscript document
IAC-05-B1.P.09.pdf (🔒 authorized access only).
To get the manuscript, please contact IAF Secretariat.