Binarization Processing Method of Adaptive Text Image Based on Pixel Neighborhood Features

Abstract

The invention discloses a binarization processing method for an adaptive text image based on pixel neighborhood features by using global and local thresholds. First, the global brightness adjustment is performed on the text image to increase the proportion of darker parts in the image and improve the brightness of the text. The gray contrast between the target character and the background in the image; then, the bicubic interpolation algorithm is used to scale the text image to increase the gap between the character strokes; then, the stroke width d of the target character is counted to determine the neighborhood calculation template Then, according to the size of the determined field calculation template, the text information of the text image is divided into text blocks; finally, the method of combining global and local thresholds is used to perform point-by-point binarization on each text block. The invention can effectively separate the character strokes from the background, avoid broken strokes, artifacts and other phenomena, maintain the connectivity of the strokes, reduce the calculation of local thresholds for pixel points in the text image, and greatly improve the calculation speed.

Description

technical field [0001] The invention relates to a method for binarizing a character image, in particular to a method for binarizing an adaptive character image based on pixel neighborhood features by using global and local thresholds. Background technique [0002] Binarization is a basic technology in digital image processing technology, and it is also a preprocessing technology for many image processing technologies. It is used in image processing such as automatic target recognition (ATR), image analysis, text enhancement and optical character recognition (OCR). be widely used. Most of the existing binarization methods belong to the thresholding method, and in different applications, the selection of the threshold determines the retention of image feature information. Therefore, the method of automatic threshold selection is worth studying. A good automatic threshold selection method can not only preserve the useful information in the image, but also reduce the time cost....

AI Technical Summary

Problems solved by technology

Dynamic threshold binarization can deal with poor-quality images or even unimodal histograms, but because dynamic thresholding methods often need to calculate the threshold for each pixel in the image, that is, to find a threshold surface for the entire image (usually Surface) has a large amount of calculation, and the calculation speed is generally relatively slow. Due to the shortcomings of time-consuming and some distortion, this hinders its development to a certain extent.

Images