ABSTRACT
This chapter describes methodologies for carrying out damage detection using both color and texture approaches and provides some insight into what technique to use in various situations and for the various types of damage encountered in the marine environment. The color based method adopts a multi-phase segmentation methodology that incorporates features from three standard image processing and data analysis techniques. The chapter evaluates a texture segmentation approach to detect and classify surface damage on infrastructure elements. It looks at the performance of the color based method alongside a texture analysis based technique, in order to give an indication of the performance levels that can be expected when a range of damage forms and surfaces are under consideration. The texture analysis involves two main stages. The first stage computes a texture characteristics map from an input color image, and the second stage subsequently classifies pixels as being either damaged or undamaged using support vector machines.
