The FPGA module consists of three sub-modules working in a pipeline:

• Adaptive binarization engine.

• Bi-tonal image denoiser / despeckler.

• CCITT Fax Group 4 encoder.

Overall module performance is between 70 and 100 mega pixel per second. Output to input compression ratio varies in a range from 40:1 to 500:1, depending how busy the image content is.

 Adaptive binarization sub-module is a highly customizable unit transforming grayscale image into a bi-tonal one. There are two adjustable absolute thresholds for unconditional conversion into black or white. Every pixel with a value in between those thresholds gets through a 7x7 window neighborhood comparison and contrast enhancement with an additional filtering of small pixel differences (adjustable tolerance). All together - the binarization sub-module is able to produce clear, crisp and distinct bi-tonal images from even very weak, dirty input documents like bent thermal-printer copies; or old, worn-out documents.

The denoiser / despeckle module removes any unwanted, remaining spots generated from the background texture. For the best control over the process, sensitivity of the despeckling filter is adjustable separately for the black spots on the white background and separately for the white spots on the black background.

The lossless CCITT Fax Group 4 image encoder module compresses bitonal images - according to the telecommunication standard: "ITU-T T.6 MMR". The encoder uses all standard options to achieve best compression ratio - usually above 30:1 for standard text-based documents (comparing to uncompressed bi-tonal, or 240:1 comparing with the grayscale module input).