Aligning the background brightness of images of text characters using a high-pass filter

Abstract

A computational method for high-frequency filtering of images with compensation for the background subsidence around images of text characters is proposed in the paper. The computational method is based on iterative filtering by the high-frequency gauss filter. The method implementation for the images where brightness of text character pixels is higher than the brightness of background pixels as well as for the images where brightness of text character pixels is lower than the brightness of background pixels is presented. Quality indicators for the background alignment of text character images have been introduced and the preliminary analysis of the proposed method has been done according to the entered quality indicators.