I'm working on OCR processing for image-based PDF files using the Google Vision API in a Python 3.11.4 environment. The documents are structured as tables, and I need to extract the text from each cell accurately. However, I'm running into an issue where a single cell containing multi-line text is getting split into separate text blocks based on the Y coordinate. This causes the order of the content to be misaligned during reconstruction. Desired Output: Manufactured Dataline_1 Address Dataline_2 Dataline_3 Current OCR Result: Dataline_2 MAnufactured Dataline_1 Address Dataline_3 How can I effectively reconstruct multi-line text in a single cell when splits it into different blocks? Any example approaches, post-processing strategies, or best practices would be greatly appreciated. Thanks Regards enter image description here Continue reading...
