Author

Di BaiFollow

Abstract

Powder-based 3D printed products are composed of fine particles. The structure formed by the particles in the powder is expected to affect the performance of the final products constructed from them (Finney, 1970; Dinsmore, 2001; Chang, 2015; Patil, 2015). A prior study done by Patil (2015) demonstrated a method for determining the centroids and radii of spherical particles and consequently reconstructed the structure formed by the particles. Patil’s method used a Confocal Laser Scanning Microscope to capture a stack of cross-sections of fluorescent toner particles and Matlab image analysis tools to determine the particle centroid positions and radii. Patil identified each particle centroid’s XY coordinates and particle radius layer by layer, called “frame-by-frame” method; where the Z-position of the particle centroid was estimated by comparing the radius change at different layers. This thesis extends Patil’s work by automatically locating particle centroids in 3D space.

The researcher built an algorithm, named “3D particle sighting method,” for processing the same stacks of two-dimensional images that Patil used. The algorithm at first, created a three-dimensional image matrix and then processed it by convolving with a 3D kernel to locate local maxima, which pinpointed the centroid locations of the particles. This method treated the stack of images as a 3D image matrix and the convolution operation automatically located the particle centroids. By treating Patil’s results as the ground truth, the results revealed that the average delta distance between the particle centroids identified through Patil’s method and the automated method was 1.02 microns (+/- 0.93 microns). Since the diameter of the particles is around 10 microns, this error is small compared to the size of the particles, and the results of the 3D particle sighting method are acceptable. In addition, this automated method need 1/5 of the processing time compared to Patil’s frame-by-frame method.

Publication Date

8-2017

Document Type

Thesis

Student Type

Graduate

Degree Name

Print Media (MS)

Department, Program, or Center

School of Media Sciences (CIAS)

Advisor

Shu Chang

Advisor/Committee Member

Elena Fedorovskaya

Campus

RIT – Main Campus

Share

COinS