Image Borders Detection Based on the Weight Model Analysis and the Morphological Gradient Operation

Abstract

The paper describes an approach to the image borders detection that is based on the use of the weight model and the operation of the morphological gradient calculating. The proposed method relates to multiscale image segmentation methods that use the Haar orthogonal wavelet transform. The border detection procedure includes smoothing the original image, building the weight model of the image (weight image), smoothing the weight image, setting the threshold weight value, performing threshold transformation of the weight image to form a binary image, and performing gradient morphological processing of the binary image. The described method allows segmenting the original image to extract information about the borders. This information is used to calculate attributes and their analysis in computer vision systems.