I believe in the Java example you should be applying .getBytes("UTF-8") to the plaintext string; not to the digest. This is because Java strings are UTF-16.

The following two programs both seem to output the same MD5 sum...

use Digest::MD5 'md5_hex';
use Math::BigInt;

my $plain  = "abcd1234";
my $digest = Math::BigInt::->from_hex(md5_hex $plain);
print $digest, "\n";
[download]

(Run the above using perl hash.pl.)

import java.security.MessageDigest;
import java.math.BigInteger;

public class Hash
{
    public static void main( String[] args ) throws Exception
    {
        MessageDigest md5    = MessageDigest.getInstance("MD5");
        String        plain  = "abcd1234";
        BigInteger    digest = new BigInteger(1, md5.digest(plain.getB
+ytes("UTF-8")));
        
        System.out.println( digest.abs() );
    }
}
[download]

(Run the above using javac Hash.java && java -cp . Hash.)