Improvements: 

• [docx] Improved the detection of standard office bar charts and variants. 
• [docx] Improved the detection of Chinese language.  
• [docx] Improved the optical character recognition preprocessing of vector text.  
• [docx] Improved the column detection of left to right aligned text.  
• [docx] Improved the stability of graphic color detection.  
• [docx] Improved the detection of header content.  
• [docx] Improved detection of white text located on a dark background.   
• [docx] Improved handling of text where the text and background colour match.  
• [docx] Improved table detection.  
• [docx] Improved the detection of diagrams.  
• [docx] Improved detection of black text located on a grey background.    
• [office] Improved language and page orientation detection. 

Bugfixes: 

• [docx] Fixed an issue causing Latin characters in a Chinese document to be misplaced. 
• [docx] Fixed an issue where a large graphic element caused text recovery failure. 
• [pdf] Fixed an issue preventing the marked property from being retained. 
• [docx] Fixed an issue preventing the detection of the correct bounds of a graphic element. 
• [docx] Fixed a performance issue where dense vector graphics prevented successful optical character recognition of a file. 
• [docx] Fixed an issue causing conversion delay of complex one-page document. 
• [docx] Fixed a bug preventing the rendering of the first page of a detected Table of Contents.