Exploratory Study of Drone Data Stabilization with Implications in Vibration-based Structural Health Monitoring

Abstract

This exploratory study presents stabilization approaches of drone-based videos with implications in vibration-structural health monitoring applications. Unlike other standalone visionbased systems such as cameras placed on a tripod, drone drifting may occur due to airborne missions especially when operated indoors or in lower altitudes. It affects the captured frames and even the slightest movement of the drone camera will reduce the data accuracy. Post-processing methods using Computer Vision (CV) and signal-processing algorithms are used in this study to explore their effectiveness and accuracy in measuring the dynamic vibration of a structure. The object study is an aluminum bar subjected to two sinusoidal vibrations then recorded using a camera embedded on a quadcopter. After video acquisition, the post-processing is started by image enhancement and scaling procedures, followed by Scale Invariant Transform (SIFT) feature detection, extraction, and matching. The scaling factor is used to convert the image coordinate to an object coordinate system before computing the displacement of the object. Data stabilization techniques are implemented in this study, first is the background subtraction method to eliminate signal drifting and second is cleaning the data from any trends. The accuracy of the proposed framework is tested by comparing the structural dynamic responses from drone measurement to a reference sensor. The Autoregressive (AR) model generated the Power Spectral Density of the signal is also compared to the measurement from a reference camera. The results show the high accuracy of the proposed method which is up to 97.52% on the dynamic response with less effect on the signal PSD. Overall, the exploratory study obtains satisfactory results and provides a new alternative to an intelligent system in the structural health monitoring field.

Chicken Diseases Detection and Classification Based on Fecal Images Using EfficientNetB7 Model

“…EfficientNetB7 is a deep learning model that has achieved better performance on image classification and recognition work. The model is frequently scaled by arbitrarily increasing the CNN 20) width, depth, or input image resolution. Even though manual modification is time-consuming and laborintensive, this method occasionally yields poor performance.…”

Improved Blob-Based Feature Detection and Refined Matching Algorithms for Seismic Structural Health Monitoring of Bridges Using a Vision-Based Sensor System

“…The global response of bridges recorded at the bent, deck, base, and abutment levels generate valuable data for estimating permanent damage to bridges. Moreover, an intelligent robotic system provided by an unmanned aerial vehicle (UAV) or drone, which is also integrated with camera technology, is capable of recording bridges and other infrastructure during earthquakes or any vibration events; it does so without disturbances from ground movement due to its airborne operation [6][7][8][9][10].…”