ABSTRACT
This chapter presents both nonphenomenological and phenomenological algorithms for nondestructive evaluation data fusion. In particular, the section will focus on two types of sensors, namely, eddy current and ultrasonic transducers. System-based methods for data fusion can be classified into four categories, namely, signal level, pixel level, feature level, and symbol level. Signal level fusion methods are usually applicable when the sensors have identical or similar characteristics, or when the relationship between the signals from different sensors is explicitly known. The morphological closing operation is capable of filling “holes” that are generally smaller than the predetermined structuring element. The closing operation can be used, therefore, to merge the unidentified defect regions. The characteristics and nature of data generated by sensors are typically different, and consequently, information must first be mapped to a common “format”. For fusing data from diffusion and wave propagation sensors, the information has to be appropriately transformed and processed before it is integrated.
