Xref: news.nzbot.com alt.binaries.pictures.comics:5838
Path: news.nzbot.com!peer02.ams1!peer.ams1.xlned.com!news.xlned.com!peer02.am4!peer.am4.highwinds-media.com!peer01.iad!feed-me.highwinds-media.com!news.highwinds-media.com!post01.iad!fx01.iad.POSTED!not-for-mail
From: astrawebfan <astra@web.fan>
Newsgroups: alt.binaries.pictures.comics
Message-ID: <2019031909261898184-astra@webfan>
References: <2019031820524659544-symm@nowhereorg>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8; format=flowed
Content-Transfer-Encoding: 8bit
Subject: Re: Testing image stats analysis - "image_size_analysis.jpg" yEnc
User-Agent: Unison/2.1.10
Lines: 21
X-Complaints-To: http://abuse.usenetxs.com
NNTP-Posting-Date: Tue, 19 Mar 2019 13:26:18 UTC
Date: Tue, 19 Mar 2019 09:26:18 -0400
X-Received-Bytes: 1658
X-Received-Body-CRC: 2451053520
On 2019-03-19 03:52:46 +0000, symm said:
> Initially I was going to use a 10% tolerance, but the very first test
> has a two-page image with height more than 10% smaller than the median
> height. But width within 5% of double the median width. ¯\_(ツ)_/¯
>
> ...0002.jpg is a wrap-around cover but there's no guarantee that an
> inside two-page spread will not be cropped/resized differently than the
> single page images.
>
> May have to go with more rigorous statistical analysis (iqr/standard
> deviations).
>
> ...0008.jpg is manually manipulated as there were no rotated images,
> but it should be representative.
Rotated pages would have an area (HxW) like the others but each
dimension would be way off, so logic like area < 1 standard deviation
and height > 1 sd. That should differentiate them from two-page spreads
and random smaller or larger images.
|
|
![Favorite](javascript:myfavor('http://nzbot.com/article/alt.binaries.pictures.comics/5838h', 'Re: Testing image stats analysis - "image_size_analysis.jpg" yEnc')){kind=link}