The Data Compression News Blog

All about the most recent compression techniques, algorithms, patents, products, tools and events.

Subscribe

Posts: RSS Feed
Comments: RSS Feed

A Faster GeoJasPer Is Now Available

Posted by Sachin Garg on 19th October 2005 | Permanent Link

Greg Coats announced an upgrade to GeoJasper, offereing 2 to 10 times speedup for Jpeg2000 conversions. Thanks to a contribution by Kaspar.

Here is his post at jasper-discussion Yahoo! Group:

I am pleased to announce the availability of GeoJasPer version 1.2.8, which offers a significant speed up for GeoTIFF/TIFF to GeoJPEG2000/JPEG2000, and GeoJPEG2000/JPEG2000 to GeoTIFF/TIFF conversions ranging from at least 2X to more than 10X, as compared to the previous version of GeoJasPer/JasPer. The software has been extensively tested, and creates image files that are bit for bit identical to those created by the previous version of GeoJasPer/JasPer.

The GeoJasPer version 1.2.8 source code, and compiled executables, are immediately available, exclusively from my web site

http://homepage.mac.com/gregcoats/jp2.html

Greg Coats

I dont follow Jasper’s mailin list very closely, but if I am correct, this speed gain comes after this contribution by Kaspar:

subject: Making JasPer faster

Hello everyone.

Don’t ask why, but after downloading JasPer I found myself staying up far too late doing performance profiles and trying to figure out how it all worked. I found out that part of the code is quite cache inefficient, and the next day I found a way to make it much faster.

I emailed Michael Adams who is aware of the slowdown, and apparently he is doing something very similar to me in the development version of JasPer. He wasn’t able to give a date for a new public release, so until then you can download my changes here - it’s just a single file:
http://www.daugaard.org/jasper/jpc_qmfb.c

The difference is quite remarkable. A 4096×4096 image that took 30 seconds to decode now only takes about 10 seconds, and encoding is similarly faster. There’s no difference in output quality or the library’s behaviour (bar any bugs I’ve introduced); the only
difference is in how memory is accessed internally.

I hope some of you will find this useful while we wait for the next public release. :-)

Kind regards,
Kaspar

For further details, you will have to read the discusison at jasper-discussion Yahoo! Group.