Image Detection Using Edge Extraction Algorithms
Abstract
In this report, we investigate alternative time-domain image detection techniques that do not require statistical models. Two distinct methods are presented; multistage pulse code modulation (MPCM) and relative entropy-based image thresholding technique. The idea of MPCM is to expand the commonly used Pulse Code Modulation (PCM) to multistage quantization to decompose an image into multiple stages. Each stage represents an image description with different details. By virtue of this decomposition, MPCM can be regarded as a progressive edge detection technique and a top-down gray-level triangle method, respectively. For vapor cloud detection, we are often interested in the detection of clouds. Without reconstructing the original image, progressive edge detection extracts edges of objects progressively from the most significant contours to the least important boundaries. The relative entropy-based approach adopts a different philosophy. It utilizes a co- occurrence matrix to efficiently segment objects from the background. It chooses an information theoretic criterion relative entropy as a thresholding technique which turns out to be a better approach than existing entropic thresholding methods in most test images.
Document Details
- Document Type
- Technical Report
- Publication Date
- Aug 01, 1993
- Accession Number
- ADA272439
Entities
People
- Chein-i. Chang
Organizations
- University of Maryland, Baltimore