| 摘要 |
A method for image edge enhancement consists in image registration, smoothing by convolution with Gaussian and calculation of image gradient in vertical and horizontal directions by means of one of first order differential operators, repeated smoothing followed by discrimination of points of intersection of boundary curves of input image obtained as result of repeated smoothing. In order to eliminate false contours a gradient image obtained by calculation of image gradient in vertical and horizontal directions by means of one of first order differential operators is superimposed into obtained image, and thinned image is obtained. After obtaining thinned image a binarization threshold is determined, pixel-by-pixel transformation on the basis of determined binarization threshold and contouring on the basis of pixel-by-pixel transformation are performed. |
