You seem to need to first index the names by the ids for later retrieval. After populating the new array, you also want to add items that weren't used for replacement (JEAN here). So, we need to remember which ones were used.

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

my @data = map {
    my %h;
    @h{qw{ ID Name Number CODE }} = @$_;
    \%h
} ['1212', 'JOE',   'XY1',  '6'],
  ['1212', '',      'XY1', '10'],
  ['4456', 'MARIA', 'TYX',  '6'],
  ['4456', '',      'TYX', '10'],
  ['8765', 'JEAN',  'HPO',  '6'];

my %name;
for my $d (@data) {
    $name{ $d->{ID} } = $d->{Name} if length $d->{Name};
}

my @new;
my %used;
for my $d (grep ! length $_->{Name}, @data) {
    push @new, { %$d, Name => $name{ $d->{ID} } };
    undef $used{ $name{ $d->{ID} } };
}

for my $d (@data) {
    push @new, $d if length $d->{Name} && ! exists $used{ $d->{Name} }
+;
}

use Data::Dumper; print Dumper \@new;
[download]