Morse has 63 total symbols. If you did a naive translation of Morse into a binary representation, you'd need 6 pixels per symbol, and of course you'd only get capital letters from the basic Latin alphabet. In Morse with actual dots and dashes, you would need at least two pixels for a dot and three for a dash, plus an additional white pixel for the space between symbols. The longest symbol is zero, which is five dashes, so you need 31 pixels to encode it. Double that if you want to put blank lines between lines of Morse code.

If you encode your data in UTF-8, you can represent every human language and a lot of supplementary symbols as well. It's based on bytes, so groups of eight pixels. Western languages only need one byte for almost every character. Other languages need more, with up to five bytes for some characters, but since we're comparing it to Morse code, I think assuming one byte per character is more than fair.

In binary, it's also a straightforward to use LZ encoding (the type of compression that zip files use). This will reduce the number of pixels you need by 60% to 80% for English language text, so if you use compressed UTF-8, you're looking at an average of two to three pixels per letter in the English portion of the text. That completely blows Morse code out of the water. The only letters in Morse that are that short are E and T, so basically binary gives you all human alphabets in the same number of pixels that it takes for Morse to give you just the letters E and T.