Yunmai's Image Preprocessing SDK allows you to process a digital document image before starting the recognition. The goal is to improve the lisibility of the text on the image to increase the recognition rate, as well as to compress the image to enhance the speed of the recognition.
The position of the document can be detected automatically and the picture cropped accordingly. This part can also deskew the image.
The orientation of the text within the document is checked and the image is automatically turned (90, 180 or 270 degrees), so that the text is horizontally positioned.
You can apply filters on the images (Black&White, Greyscale, enhanced contrast) to clean the image before recognition.
The image's background is cleaned of the potential noises to present the user a clearer image and to remove artifacts that can affect recognition.
The image's contrast is adjusted to provide a better looking document picture as well as to help the OCR engine to differenciate text from background.
The image is compressed to increase the speed of recognition and to reduce the volume of data for transfer and archive.