For your first requirement, a regex is probably safe and effective, since (unless I'm having a Sr. moment) the html 4.x standard does not allow an image tag with a literal ">" inside the tag.

One way to approach the job, therefore, is to extend your regex with less-greedy (aka "minimally greedy") matching and a lookahead. Here's a sketch, minus file-handling, CGI, etc:

#!/usr/bin/perl
use strict;
use warnings;
#825146

my @line = <DATA>;
for my $line(@line) {
    chomp $line;
    if ( $line =~ m/(<img .*?[^>]+)/  ){
        print  "<g:image_link> " .  $1 . "> </g:image_link>\n";
    } else {
        print "\t nope: $line \n"; # you may want to send this to a di
+fferent file
    }
}

__DATA__
<p><img src="http://www.mysite/graphics/blue.jpg" alt="Hey" width="100
+" height="100" ><br>yada yada</p>
<p><img src="../grapics/blue1.gif" alt="Yo" width="200" height="75"></
+p>
<p>foobar with no img</p>
<blockquote><img width="75" height="75" src="blue2.png"></blockquote>
[download]

Output:

<g:image_link> <img src="http://www.mysite/graphics/blue.jpg" alt="Hey
+" width="100" height="100" > </g:image_link>
<g:image_link> <img src="../grapics/blue1.gif" alt="Yo" width="200" he
+ight="75"> </g:image_link>
     nope: <p>foobar with no img</p> 
<g:image_link> <img width="75" height="75" src="blue2.png"> </g:image_
+link>
[download]

BUT take the advice from pemungkah above: Use a parser! Trying to deal with all the possible unwanted tags in a form with regexen is going to get you deeper and deeper into complexities.

And if you're planning to read user input from a form, for heaven's sake, read about untainting. You really don't want to let the fumble-fingered or malicious run around loose in your playground.