can i use my own segmentation result to replace tesseracts'?

[陈袁沁]

I find tesseract-ocr's segmentation is not good for some images.

So i do the segmentation by myself. 

I have tried "setRectangle", but it will recognize character by character. And the result is worse than sending the whole image to ocr engine.

I guess tesseract's post-procesing have fix many error result with context information.

So how can i just replace all the segmentation of Tesseract? something like set the "block_list"?