improve image paring

[aoout]

[aoout]

<a class="duokan-footnote" epub:type="noteref" href="#note1" id="note_ref1"><sup><img alt="note" class="footnote" src="../Images/note.png"/></sup></a>

    <aside epub:type="footnote" id="note1">
      <a href="#note_ref1"></a>
      <ol class="duokan-footnote-content" style="list-style:none">
        <li class="duokan-footnote-item" id="note1" value="1">注：日本棒球队</li>
      </ol>
    </aside>

[aoout]

[aoout]

<img class="inline" alt="" src="Image00003.jpg" />

[aoout]

Another solution is to convert the format of the pictures near to the text.

[aoout]

Another solution is to convert the format of the pictures near to the text.

After thinking about it carefully, this is a very bad approach. It seems I need other ideas.

[aoout]

就算不能兼顾所有，起码先把那些用图片表示的中文生僻字的解析研究一下。

[aoout]

Partially resolved by https://github.com/aoout/obsidian-epub-importer/issues/99