An approach using mathematical morphology and support vector machines to detect features in pipe images

Abstract

This paper presents a new approach to detecting features in pipe images based on a generalisation of the erosion operation. The pipe images can be segmented using support vector machine or other method. The binary image obtained in this way contains a principal connected component made up from the pipe flow lines, the pipe joints and adjoining defects. The morphological analysis allows the principal component of the segmented image to be decomposed into its components. Generalisations of the dilation and erosion operations called-dilation and -erosion are defined. Some simple properties of these operations are derived. © 2008 IEEE.