PDF24’s Force OCR option currently doesn’t replace or re-recognize existing text layers — it only adds OCR if none is found. To improve your file:
First, remove the old text layer (e.g. print to image-PDF, or use tools like Ghostscript/PDF-XChange to rasterize).
Then run OCR in PDF24 again — this time it will process the clean image and generate a new text layer.
That way, you bypass the scanner’s poor OCR and let PDF24 handle recognition from scratch.