The regex version is nice. This is more explicit:

sub scan_seq {
    my ($len, $seq) = @_;
    my %sub_seqs;

    if ($len > 0)
    {
        ++$sub_seqs{ substr($seq, $_, $len) }
            for 0 .. length($seq) - $len;
    }

    return \%sub_seqs;
}
[download]

Update: Changed local our to my.