Comments on: Open and Shut [Equation of the Week] http://groundtruth.info/AstroStat/slog/2008/eotw-open-and-shut/ Weaving together Astronomy+Statistics+Computer Science+Engineering+Intrumentation, far beyond the growing borders Fri, 12 Aug 2011 17:44:19 +0000 http://wordpress.org/?v=2.9 hourly 1 By: hlee http://groundtruth.info/AstroStat/slog/2008/eotw-open-and-shut/comment-page-1/#comment-259 hlee Thu, 26 Jun 2008 02:00:39 +0000 http://groundtruth.info/AstroStat/slog/?p=326#comment-259 I have the same awe as in your, <i>...it is rather surprising how little use astronomy makes of the vast body of work on image analysis...</i> I guess it's due to different objectives in image analysis. Astronomers consider image analysis as data preprocessing for science. For computer vision people, image analysis itself is the main goal; cleaning images (noise removal), compressing images (noiseless coding and transmission), orthogonalize components (screening anomalies), and so on. There's no physics in images unless known (such as poisson photon noise instead of gaussian). It's pity that such vast reservoir of algorithms by mathematicians and computer scientists is of little use for error bar calculation. By the way, the order in dilate and erode operators reminds me the process of compressed sensing. I was curious about any differences caused by random masking when I wrote <a href="http://groundtruth.info/AstroStat/slog/2007/compressed-sensing-and-a-blog/" rel="nofollow">compressed sensing and a blog</a>. (No will to verify but from the flow chart of compressed sensing, masking could happen before or after photon read-ins.) I have the same awe as in your, …it is rather surprising how little use astronomy makes of the vast body of work on image analysis… I guess it’s due to different objectives in image analysis. Astronomers consider image analysis as data preprocessing for science. For computer vision people, image analysis itself is the main goal; cleaning images (noise removal), compressing images (noiseless coding and transmission), orthogonalize components (screening anomalies), and so on. There’s no physics in images unless known (such as poisson photon noise instead of gaussian). It’s pity that such vast reservoir of algorithms by mathematicians and computer scientists is of little use for error bar calculation.

By the way, the order in dilate and erode operators reminds me the process of compressed sensing. I was curious about any differences caused by random masking when I wrote compressed sensing and a blog. (No will to verify but from the flow chart of compressed sensing, masking could happen before or after photon read-ins.)

]]>