Masters Theses


"A computer vision approach is investigated which has low computational complexity and which compares near-infrared and visible image systems. The target application is a surveillance system for pedestrian and vehicular traffic. Near-infrared light has potential benefits including non-visible illumination requirements. Image-processing and intelligent classification algorithms for monitoring pedestrians are implemented in outdoor and indoor environments with frequent traffic. The image set collected consists of persons walking in the presence of foreground as well as background objects at different times during the day. Image sets with nonperson objects, e.g. bicycles and vehicles, are also considered. The complex, cluttered environments are highly variable, e.g. shadows and moving foliage. The system performance for near-infrared images is compared to that of traditional visible images. The approach consists of thresholding an image and creating a silhouette of new objects in the scene. Filtering is used to eliminate noise. Twenty-four features are calculated by MATLAB♭ code for each identified object. These features are analyzed for usefulness in object discrimination. Minimal combinations of features are proposed and explored for effective automated discrimination. Features were used to train and test a variety of classification architectures. The results show that the algorithm can effectively manipulate near-infrared images and that effective object classification is possible even in the presence of system noise and environmental clutter. The potential for automated surveillance based on near-infrared imaging and automated feature processing are discussed"--Abstract, page iii.


Watkins, Steve Eugene, 1960-

Committee Member(s)

Wunsch, Donald C.
Stanley, R. Joe


Electrical and Computer Engineering

Degree Name

M.S. in Computer Engineering


Missouri University of Science and Technology

Publication Date

Summer 2012


xiii, 150 pages

Note about bibliography

Includes bibliographical references (pages 21-23).


© 2012 Kathryn Nicole Rodhouse, All rights reserved.

Document Type

Thesis - Open Access

File Type




Subject Headings

Image processing
Infrared imaging
Neural networks (Computer science)
Surveillance detection

Thesis Number

T 10069

Print OCLC #


Electronic OCLC #