MathGroup Archive 2011

[Date Index] [Thread Index] [Author Index]

Search the Archive

Re: bug

  • To: mathgroup at smc.vnet.net
  • Subject: [mg123259] Re: bug
  • From: Matthias Odisio <matthias at wolfram.com>
  • Date: Tue, 29 Nov 2011 07:05:57 -0500 (EST)
  • Delivered-to: l-mathgroup@mail-archive0.wolfram.com
  • References: <201111241154.GAA28922@smc.vnet.net>

On 11/24/11 5:54 AM, Szabolcs wrote:
> It appears that the example at [1] does not work in version 8.0.4 on Windows.  Could someone please test this, and confirm if one of the 8.0.x broke it?  Please also mention the precise version you have.  The code can be copied and pasted verbatim, and the correct output (two slanted lines) is shown on the linked page.
> 
> The difference may be due to a different output of the Radon[] function.
> 
> [1] http://dsp.stackexchange.com/q/682/581
> 


The coordinate system used in the Hough transform (Radon[img,
Method->"Hough"]) changed after M8.0.0 to match the one of Radon
transform. This in turn has changed the behavior of ImageLines.

For the specific example above the automatic settings look more
pleasant with M8.0.0 than with M8.0.4; However I believe M8.0.4 is
overall better at detecting lines.

Matthias Odisio
Wolfram Research



  • Prev by Date: Re: Problem with Patterns and Integrate
  • Next by Date: Re: import data problem
  • Previous by thread: Possible Radon[] bug
  • Next by thread: ContourPlot ColorFunction workaround