James Vences, “Hough Cylindrical Transformation to Detect and Quantify Escherichia coli Cells Within a Photo”
Mentor: Marcia Silva and Dominic Marr, Global Water Center
Poster #133

The purpose of this research project is to automate a program that counts the amount of Escherichia coli cells within a photo. The importance of developing a program that could detect and quantify the amount of E. coli cells is crucial towards the development of a sensor that can take photos of a water sample and quantify E. coli cells in real time. Developing an efficient program and using it for a sensor will drastically increase the speed at which the information gathered can be quantified. Within the field of computer vision, one of the challenges of automating digital image analysis is shape detection. From extensive research scholars in computer vision have derived the Hough Transformation, a technique in image analysis that solves challenges within computer vision such as background noise, extraneous data such as unnecessary or extra edges, and incomplete data when performing boundary detection of shapes. The Hough Transformation was originally used to find lines within a photo but in 1972 it was used to find shapes such as circles. An E. coli cell has a cylindrical shape with a radius of 0.5 micrometers and a length ranging from 1.0 to 2.0 micrometers long. To quantify E. coli cells Hough Transformation will be applied to find cylindrical shapes within a water sample photo. The current state of the Hough transformation algorithm takes up 5D Hough space. An acceptable range is 3D Hough space which refers to the cylinder being three-dimensional. Unknown parameters such as orientation and edge location of E. coli cells increments the Hough Space. A high Hough space makes the program inefficient due to the high run time and increased space complexity making the program impractical. A proposed solution to this problem is to combine 2D and 3D Hough Transformations to address these unknown parameters.