Additionally, if the aim is to make the editorial text searchable, it is only really necessary to correct likely keywords that might be searched for. Given the size and complexity of a newspaper page, and the fact that part of the content consists of advertisements, which might ideally be made searchable but are less important than the editorial text, it would probably be quicker and more practical to simply zone the editorial text areas and recognise them. Since our goal is to simplify the batch resizing and compression of masses of images, add the Resize action to the active script with a click on Add action -> Image -> Resize. I presume you want to make the newspaper text searchable and then save each page as a PDF file using FineReader's 'Text under page image' mode? I think your problem is really a question of strategy: what you want to achieve and the most practical way to achieve it. I have tried many different combinations of nconvert.exe command line arguments. The image enhancement that might possibly improve the recognition results might be to resample the page image to increase the resolution, but the image dimensions are already quite large and the reading results are probably limited more by the accuracy of individual printed characters than by image resolution. Using XnConvert it worked perfectly the first time, just selected TIFF as the output. 1 Add Files Clicking the Add Files button will aid the user in adding the photos he or she wants to convert. I've now opened the image in FineReader 12 and find the text recognition surprisingly good considering the resolution and discolouration of the source image. I established early on, or possibly rediscovered, that NConvert doesnt seem to accept input with a wildcard in the path as well as in the filename, C:X//.png for example.
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |