A Comparison between Pixel Intensity and Texture Analysis based Damage Detection Techniques

Imaging based damage detection techniques are increasingly being utilized alongside traditional visual inspection methods to provide owners/operators of infrastructure with an efficient source of quantitative information for ensuring their continued safe and economic operation. However, choosing a suitable damage detection technique that will provide a high level of quantitative information is often a timely and challenging prospect as there exists a wide array of algorithms currently available. These algorithms may be partitioned into one of two groups; pixel intensity based methods and texture analysis based methods. The algorithms in each group are naturally suited to different applications, depending largely on whether the damaged region under consideration is more separable from the background based on colour or on texture. This paper compares two algorithms, one from each category, which have previously been proposed in the domain of Infrastructure Maintenance Management (IMM). The algorithms are applied to a range of scenes featuring various forms of damage and their performance is investigated for best detection on the basis of performance coordinates in the Receiver Operating Characteristic (ROC) space. The general superiority of the pixel intensity based approach over the texture analysis based approach is demonstrated, in particular when considering High Dynamic Range (HDR) imagery