Output places image tag before the title tag and wraps both image tag and title tag in an abstract element in dita output; which appears to be illegal for topic type element.
DITA is intended to be semantic, not presentational. If you want a graphic by the topic title in DITA, you add it during output processing. It does not belong in the DITA itself. So you need to mark it for deletion. How you do that depends on how it was added in Frame. If it is a FrameAbove, it's simple, per User's Guide par. 21.3.8, "Keeping or removing reference frames":
[HTMLOptions]
; RemoveFramesAbove = No (default) or Yes (can be overridden)
RemoveFramesAbove=Yes
removes all of them, which is probably what you want. If it's anchored in the text, apply a condition and exclude that condition when converting to DITA.
Presentational elements in DITA
DITA is intended to be semantic, not presentational. If you want a graphic by the topic title in DITA, you add it during output processing. It does not belong in the DITA itself. So you need to mark it for deletion. How you do that depends on how it was added in Frame. If it is a FrameAbove, it's simple, per User's Guide par. 21.3.8, "Keeping or removing reference frames":
[HTMLOptions]
; RemoveFramesAbove = No (default) or Yes (can be overridden)
RemoveFramesAbove=Yes
removes all of them, which is probably what you want. If it's anchored in the text, apply a condition and exclude that condition when converting to DITA.
HTH!