Your image hash parsing code makes the regex engine work harder than necessary. I've reworked this bit to show several alternatives.

The problem is that /.* .*/ creates a lot of unnecessary backtracking. The regex engine is greedy, so it will scan the whole string and match it to the leading '.*'. Then it has to back track and remove one character at a time until it finds a space. If you write /[^ ]* .*/, then the engine can read until it finds a space, then move on the next match criterion, must be a space, match that and slurp in the rest of the string to match the final '.*'. I once read a great article that talked about this in detail, but I can't track it down.

my %image_hash = ( );

while (<IMAGEHASH>) {
    # Limited scope of $search/$replace
    # You could just use split to split on whitespace:
    my ($search, $replace) = split;

    # Or if you want to split on the first single space character, and
+ force only 2 items:
    # my ($search, $replace) = split ' ', $_, 2;

    # or you could use a regex:
    # my ($search, $replace) = m/([^ ]*) (.*)/);

    # Suppress uninitialized warnings.
    # check for non-zero length, 
    # although eq '' checks should be fine, too.
    {   no warnings 'uninitialized';
        if ( length $search and  length $replace ){
            warn "Found uplicate entry for: $search\n"
                if exists $image_hash{$searc};

            $image_hash{$search} = $replace;
        }
    }
}
[download]

Warning: all code untested.