Abstract: Currently, image processing methods, including methods of boundary detection in digital images, which are often used in technical vision systems to perform further recognition, do not lose their relevance. This article discusses different approaches to configure settings which are used to modify Canny detector for the allocation of boundaries based on the model of energy characteristics of wavelet transform. Main stages of this modification are considered and their description is given. For such stages as the formation of gradient direction’s image and double threshold filtering, the analysis of possible approaches to calculating the required parameters is performed. In addition, presented image processing algorithm is applied to standard test images and analysis of acquired results is carried out. As a result, an approach based on the calculation of ratio of the values of vertical and horizontal detailing coefficients obtained during the wavelet transform, is proposed for the processing of digital images by means of modified Canny algorithm to form an image of gradient directions. It is proposed to use an approach based on the calculation of the upper threshold value by means of a histogram of the gradient direction matrix to calculate the parameters of the double threshold filtering stage. Obtained results can be used to configure modified Canny algorithm, which in turn allows for less time to perform the allocation of the boundaries on digital images, which can be used in real-time vision systems. Thus, the choice of the most optimal methods in terms of reducing the computational and time costs for the algorithm to calculate its parameters is made. Further direction of work to improve the efficiency of presented approaches in terms of the quality of resulting image modification and options for further processing to obtain clearer boundaries are proposed.
Index terms: image processing, edge detection, wavelet analysis, energy features.

Contacts

Russia, 659305, Altai region, Biysk,
Trofimova Street, 27, room 404B
Tel. + 7-923-162-93-27
(executive secretary -
Golykh Roman Nikolayevich)
e-mail: info@s-sibsb.ru

The certificate