On a method of edge detection in digital images

The methods of digital image detection are considered. Some of them are based on calculation of an image gradient (Roberts’ cross detector); others are based on the notion of the central difference (Prewitt operator, Sobel operator and Scharr operator). There are also approaches that calculate image laplacian the advantage of which is a perfect rotational symmetry. As an alternative to those calculating image gradient and laplacian statistical methods are used. The advantages and disadvantages of these methods are mentioned. The method based on the author’s function used in the creation of a low-pass filter is proposed. The method specifically calculates image gradient suppressing the image noises at the same time. The comparison of the proposed method with the standard one revealed its advantages — better accuracy of the calculation of a gradient and noise suppression, and its disadvantages, i. e. longer period of image processing.