Computer Vision Based Displacement Measurement of Shake Table

摘要

In our surroundings, many things move from one place to another some displace significantly, others only execute a small motion. When it comes to motion estimation, it is easier to quantify large displacements whereas small movements like facial expressions etc. are relatively difficult to detect and measure. Mark Dow at University of Oregon Brain Development Laboratory in February 2009 [1] devised an algorithm to estimate complex small motions in scenes. The algorithm sums pixel wise difference between frames to estimate motion. The proposed framework uses the core of this algorithm to estimate strength of a building structure in laboratory. The framework is a computer vision based and can be used to estimate automatic variation in the building displacement before going it falls. Conventional methods use LVDT Transducers which are expensive and consume much human involvement. The proposed framework was tested using several synthetic data sets and two real data sets taken from shake table installed in an Earthquake laboratory. The algorithm is currently applied for post processing of the data obtained from laboratory however, it performs fast enough to be used for real-time monitoring of concrete structures like bridges, dams etc.