Damage imaging of structural components in the field of Non Destructive Evaluations (NDE) and Structural Health Monitoring (SHM) using ultrasonic waves is usually performed by conventional imaging techniques, such as DelayAnd-Sum (DAS), by back-propagating the recorded waveforms to identify locations and size of defects and damages. This technique results in sidelobes and artifacts that worsen the accuracy of the damage identification. Here we propose a novel imaging approach that derives from the well-known technique of Matched Field Processing (MFP), often used in underwater acoustics and seismology. In MFP, the source or damage is located by a matching procedure between measurements (“data vector”) and expected responses (“replica vectors”) computed for each point of the imaging volume. In this work, we apply this matching approach only to selected features extracted from the recorded waveforms. These features, for example time-of-flights or amplitudes, will be selected for multiple modes of propagation of the ultrasonic waves (longitudinal and shear in bulk waves, multiple guided modes in waveguides). By considering multiple features and multiple wave modes, it is possible to increase the performance of this matching procedure, which can be possibly further improved by also combining different signal frequencies and excitation sources in analogy with biomedical ultrasonic imaging. A correlation metric showing high computational efficiency in the image reconstruction process will be tested as matching coefficient. Applications of this imaging approach to a metallic plate with holes and simulated defects will be shown.