Adaptive segmentation method of pressed character image based on Wellner algorithm

The image of pressed characters on the surface of metal workpieces in industry has obvious unimodal characteristics, for this feature, this paper proposes an adaptive segmentation method based on Wellner algorithm, this method is used to segment the pressed character image whose character gray value is similar to background gray value. Firstly, we use uniform illumination to capture grayscale images. Next, the Retinex algorithm is used to enhance the details of the character edge, the grayscale distribution range is expanded to improve the image contrast. Then, the bilateral filtering algorithm is used to filter the image noise. In this paper, the pixel gray value of a certain point is selected as the center, the row and column mean value of the pixel is calculated, at the same time, the mean value of the pixel gray value in the 8-connected region that it belongs to the pixel selected to be the center is calculated. The algorithm applies the “center-around” idea, the Wellner algorithm is improved with the mean value and the image pixel points are traversed to achieve image binarization. Finally, the final segmentation result is obtained by combining morphological operations. The verification experimental results show that the proposed method has good self-adaptiveness and accuracy for the gray-scale histogram image with unimodal characteristics.