Oftentimes, when using Optical Character Recognition software to extract and recognize text from an image, you get bad results if use default configuration, especially if the image scan resolution is less than 150 dpi. If recognize computer graphics, like text from screenshots or images from the web, then things become even more worst, because the standard resolution of web graphics is not more than 72 dpi. That is why need to pre-process images to increase recognition accuracy.
I created a tool, called "Image Reader", which is an online text recognition web application with image pre-processing features. It allows to upload an image with text, apply different filters to it and then extract text from it. This tool can be used as a playground to try and understand what exactly needs to do with images to increase text recognition quality. You can see a demo of this app here: https://youtu.be/WsuZHyAuDTI and of course, try the app itself.
Top comments (0)