Structural Health Monitoring (SHM) and condition assessment deal with inspecting the health and integrity of the monitored systems. Although robust damage detection methods have been proposed in the recent two decades, there are intensive ongoing investigations to tackle the practical and technical open challenges, such as smart sensing, sparse sensor measurements and real-time damage detection. This thesis is devoted to discussing pattern recognition and damage detection methods using vibration, acoustic and vision perspectives.First, a machine learning-based approach is proposed for object classification by texture analysis. Bag-of-Words (BoW) and Support Vector Machine (SVM) techniques are used to extract the features and train an identifier, respectively. The method is particularly exploited for tie/ballast image classification at Rail Defect Facility of UC San Diego by mounting a high-speed camera on a cart moving with walking speed. Second, a deterministic vibration-based method is proposed for damage quantification in the structures, using sparse sensor measurements. The estimated damages are then further tuned by repeating the proposed approach to reach more accurate results. The method is employed for damage detection in lab-scale and full-scale building structures. Defect imaging in plates using data-driven Matched Filed Processing (MFP) is the last concept discussed in this thesis. Under the Born approximation, difference between the responses of the damaged and pristine plates is computed as the data set containing the defect’s acoustic signature, and conventional and adaptive beamformers are used to perform the MFP and localize the defect. The method is employed for damage detection in an aluminum plate.