Newsgroups: sci.image.processing
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.alpha.net!uwm.edu!vixen.cso.uiuc.edu!howland.reston.ans.net!ix.netcom.com!netcom.com!perry
From: perry@netcom.com (Perry West)
Subject: Re: TV Resolution?
Message-ID: <perryD6B5t0.z2@netcom.com>
Organization: NETCOM On-line Communication Services (408 261-4700 guest)
X-Newsreader: TIN [version 1.2 PL1]
References: <3l6l75$2gm@usenet.rpi.edu>
Date: Fri, 31 Mar 1995 14:05:24 GMT
Lines: 33
Sender: perry@netcom2.netcom.com

TV resolution is the maximum number of white and black lines that can 
span the image (either horizontally or vertically) and still be 
discriminated.  In actual practice, it is judged subjectively, and proves 
to be fertile ground for specmanship (this is a general statement not 
directed toward any specific camera company).

There are 525 total scan lines in an RS170/RS330 frame.  Only between 480 
and 488 of them carry useful video information (the specifications give 
some leeway here).  So, the Pulnix camera, having 494rows of sensing 
elements, is a bit of an overkill.

Looking in the vertical direction, we know (or at least strongly expect) 
that if the image consists of only one white and one black bar, we will 
have 100% contrast.  Also, from sampled data theory (Shannon's sampling 
theorm and the Nyquist criteria) if, for a sensor with 480 rows of 
sensors, we have 480 lines (240 white and 240 black), we will have no 
detectable constrat.  The bars relate to spatial frequency (cycles/mm 
rather than temporal frequency which is cycles/sec or Hertz).  Therefore, 
the sensor can be expected to have a spatial frequency response which is 
achievable contrast as a function of cycles/mm or line (pairs) spanning 
the image sensor.  In the example above we know it goes from 1 at very 
low frequencies to 0 at a frequency which is one-half the number of 
rows.  If we study enough cases, we could draw the whole frequency 
response curve.  In optics and cameras, this frequency response curve is 
called the modulation transfer function (MTF).  When we use bars instead 
of sine waves, we get the contrast transfer function (CTF) which is close 
enough to the MTF from most practical purposes.

The TV response, then, is intended to be the point on the MTF curve where 
the spatial frequency response still gives a *useable* picture.  Useable 
is supposed to be about 10% to 15%, but it is a subjective call off a 
test pattern, and usually near the center of the image.

