Try adaptive thresholding

[falktan]

In first test the results of the OCR from TesseractJS are not very convincing. One hint on how to improve on this that I read somewhere a while ago is to use a more suitable thresholding method. The one described here might be suitable:
https://docs.opencv.org/4.5.2/d7/dd0/tutorial_js_thresholding.html

• implement a adaptive thresholding as a pre-processing step
• find appropriate parameters
• check if this really improves over previous results without pre-processing

[falktan]

Here is a document supporting that this might be a good idea and also making a couple of other suggestions:
https://tesseract-ocr.github.io/tessdoc/ImproveQuality

[falktan]

Started work to test this here:
Started work here:
https://github.com/falktan/ocrjs

[falktan]

first implementation in https://github.com/falktan/ocrjs
Any further steps are tracked through #3 or new tickets to be opened.