It is generally expected that European names be spelled correctly in academic correspondence and citations. This is expected, even when they use symbols not in the English language, when publishing in English (provided that a modern typesetting system will support it).
For instance, the German name "Müller" is expected to be spelled correctly, even though it could be spelled "Mueller" in the English Alphabet (without loss of information). Similarly, Gaelic names such as "Ó Ceallaigh" would not be Anglicised, despite a long history of this having been done before.
However, this is not the case for Asian languages, even Japanese names where they do not typically have the culture of adopting a nickname in Western Countries. For example, "田中" would be spelled as "Tanaka" to confirm with English language readers, despite Japan having it's own phonetic conventions to give the desired reading (e.g., "タナカ").
For a more comparable example, why is it ok to spell “Tokyo” in English when “Muller” is incorrect. The correct romanisation of 東京 is Tōkyō. Yet this is not used, nor is Toukyou or とうきょう which would both be more accurate. This misspelling occurs for names of people in Asian languages as well as names of places.
It's typically argued that this is because English-speaking audiences could not read Japanese names in their writing system but the same could be said for the umlaut, which is often mispronounced or misused (e.g., Mötley Crüe). With digital typesetting systems, would be entirely possible to spell "田中" correctly in a citation, even by someone who cannot understand the meaning, just as we do for diacritics for European names.
If spelling someone's name correctly is a matter of respect, when is it necessary to do so and why are there exceptions to this?