Date of Award

Summer 2014

Degree Type


Degree Name

Master of Science in Civil Engineering (MSCE)


Civil Engineering

First Advisor

Hubo Cai

Committee Chair

Hubo Cai

Committee Member 1

James S. Bethel

Committee Member 2

Phillip S. Dunston


A construction site presents a dynamic scenario. Locations of multiple objects are continuously changing and a lot of objects enter and exit the site in high frequencies. Meanwhile a construction activity consists of a large amount of stochastic operations, many uncertainties occur when making decisions. Believing that detecting, locating and tracking dynamic construction objects in real time improve construction productivity and enhance construction safety, a large number of studies have applied a variety of sensing technologies to construction sites. Hybrid image-point cloud sensing technologies, such as color-depth cameras, have a great potential in achieving real time object recognition and tracking in an efficient manner. However, there is one big challenge unsolved which is caused by the complexity and variety of construction objects — lacking a generic modeling approach to analyze varying objects, and a follow-up "seamless" workflow to automated detecting, locating and tracking dynamic objects in real-time or near real-time. This study proposes a key nodes modeling approach to facilitate the detecting, locating and tracking process. First, a supervised training process is designed based on key nodes model which can lead to a standard templates library for typical construction objects. Then the key nodes model abstracts the distinguished characteristics of object's geometric shape, meanwhile incorporates the kinematic constraints for adjacent parts of objects. Object detection is realized by comparing the real object components and segmented contour templates in a traversal order of the tree structure of the key nodes. Since color-depth cameras capture not only images, but also the distance between detected objects and the camera itself in a format of three-dimensional (3D) point clouds, the 3D spatial location of the detected key node is obtained by linking its image pixel value with corresponding 3D point coordinate. Therefore, the objects can be dynamically located and tracked in a very smooth and efficient manner. ^ The key nodes modeling approach is proved to be efficient for detecting, locating and tracking key nodes of objects, which also has great potential for interactive analysis and behavioral prediction in construction monitoring and safety management.