Separate the text from the others elements on the image

I have tried to resolve my issue by the threshold of OTSU by creating a mask of the image but I did not get a good result.

that IS a good thresholding result.

please tell, how exactly you try to extract text from it, and where your current problem is

After the processing I had this result, but I would like to minimize the deterioration of the texts.

A chuck of it:

sorry, but it’s unclear, how you get from image 1 to image 2, and what you want from either

imho, we need to start all over: what is your original image, and how do you expect to extract text from it ?
(maybe a solution does not even use thresholding ?)

Early I posted only a chuck the original image…I already update the same part of the image processed

Here it’s the original image. I would like to remove the lines and remain only with the text for the text recognition, I’ve tried text recognition with ocr, keras_ocr and others lib’s and i didn’t get good results. So I hope this approach will be a good one.