Logo image
Back
AUTOMATED DETECTION OF SWIMBLADDER REGIONS IN THE STRIPED BURRFISH USING A MODIFIED SNAKES ACTIVE CONTOUR ALGORITHM
Thesis   Open access

AUTOMATED DETECTION OF SWIMBLADDER REGIONS IN THE STRIPED BURRFISH USING A MODIFIED SNAKES ACTIVE CONTOUR ALGORITHM

Jodie Paige Gray
University of West Florida
Master of Science (MS), University of West Florida
2015

Metrics

Abstract

This thesis proposes an automated technique for the initialization stages of the snakes active contour algorithm. Recent research has been conducted for automated snakes techniques due to current limitations of the algorithm because of the requirement for manual initialization of the snake contour by a human user. The use of manual initialization in large data sets is time consuming and inefficient. Automated methods allow for faster analysis among large data sets and unbiased segmentation results on images. Testing of the automated method is performed on sets of 2D x-ray images of burrfish. An automated form of the snakes algorithm is used to detect swimbladder regions in each image and results are compared to human identified ground truth regions. Automation of the initialization stage of snakes is achieved by a combination of image thresholding, blob detection, and mathematical methods pertaining to the known anatomy of burrfish. Given an x-ray image, the algorithm is initialized by performing blob detection and only placing initial snake boundaries that correspond to accepted swimbladder shapes within desired regions of interest relative to other distinct features within the image. This method allows for the entirely automated detection and segmentation of swimbladder regions in burrfish x-ray images.
pdf
uwf:61100DownloadView
Open Access

Details

Logo image