Date of Award

Fall 2023

Access Type

Thesis - Open Access

Degree Name

Master of Science in Mechanical Engineering


Mechanical Engineering

Committee Chair

Dr. Christopher Hockley

First Committee Member

Dr. Eric Coyle

Second Committee Member

Dr. Brian Butka

College Dean

Dr. Jim Gregory


In this work, an initial prototype of a monocular camera system capable of retrieving depth-from-focus using a liquid focus-tunable lens is constructed out of hobby-grade photography equipment. This concept has been explored previously in laboratory settings using specialized equipment; this work seeks to determine the feasibility of retrieving depth-from-focus using commercially available components. To achieve this, an iterative exploration of existing techniques was performed to verify their utility in the final ensemble of processes to retrieve depth from 2D images. Initially, blurry images were simulated by applying Gaussian blur to test images to verify the functionality of a Laplacian of Gaussian-based algorithm capable of determining image clarity, a sliding gantry was then constructed to move a camera through the environment and test the image clarity algorithm on real-world data as well as test methods to create a composite image of the most in-focus pixels from a focal stack of images collected while the camera was in motion. Following this, the depth retrieval algorithm was tested on a geared lens setup in which a gear-driven fixed focal length lens was attached to a camera and driven such that the distance between the lens and the imaging sensor in the camera was varied to change the optical power of the lens. This setup suffered from several limitations but provided significant insight into the fundamental principles governing depth-from-focus retrieval. Finally, 12mm, f/6, Liquid Lens Cx Series Fixed Focal Length Lens from Edmund Optics was attached to a Raspberry Pi Global Shutter camera to retrieve depth from an environment. This lens can vary its optical power by applying a voltage to the liquid lens which can be done automatically from a microcontroller at a high rate of speed. This operated with limited success and produced a very noisy depth map and point cloud of the environment. This work concludes with suggestions for future work to significantly improve the depth retrieval functionality of the liquid lens setup.