OPTIMIZATION OF A VISION-BASED SHM SYSTEM FOR IMAGING SUBSURFACE IMPACT DAMAGE IN COMPOSITE STRUCTURES

Abstract

This paper describes the optimization of a computer vision-based technique for visualizing subsurface barely visible impact damage (BVID) in composite structures. The system uses a piezoshaker for exciting a guided wavefield with a sweeping frequency in the near-ultrasonic frequency range, a digital camera or stereo-camera for recording surface dynamics, and an energy-based damage imaging condition for identifying local resonance from standing waves within damage regions. The method for extracting surface dynamics from the videos was optimized, evolving through three techniques – 3D digital image correlation (DIC), off-axis 2D DIC, and image decorrelation – with the ultimate goal of creating a practical and efficient visionbased method for BVID inspection. Each proposed technique was applied to inspect two CFRP composite-honeycomb panels that had been subjected to low-velocity impacts. Damage images produced with all techniques for a 100-mm  100-mm FOV using a three-second video show accurate damage imaging ability. The processing time required by the optimized decorrelation technique for extracting surface dynamics is approximately 15 times less than 3D DIC and 5 times less than off-axis 2D DIC. The increased efficiency, reduced complexity, and demonstrated accuracy of the system suggests a high potential for practical baseline-free and in-time computer-vision-based structural health monitoring (SHM) and baseline-free subsurface damage imaging for critical composite structures.