Dwight D. Egbert

Review of computer vision education

Computer vision is becoming a mainstream subject of study in computer science and engineering. With the rapid explosion of multimedia and the extensive use of video and image-based communications over the World Wide Web, there is a strong demand for educating students to become knowledgeable in computer imaging and vision. The purpose of this paper is to review the status of computer vision education today.

Fuzzy ARTMAP neural network compared to linear discriminant analysis prediction of the length of hospital stay in patients with pneumonia

On a database derived from patients hospitalized with pneumonia, the authors compared the cross-validated predictions of linear discriminant analysis (LDA) to a new self-organizing supervised neural network that incorporates fuzzy set logic into adaptive resonance theory mapping (ARTMAP) to simultaneously predict outcome and define category patterns with outcomes. The purpose of this study was to determine whether such a self-organizing neural network could accurately predict the length of stay of patients admitted to a community hospital with a diagnosis of pneumonia. Unbiased proportionate reduction in error using ARTMAP was 50% greater than LDA. Under conditions of simulated noise and increasing-proportion learning, ARTMAP demonstrated further advantages over LDA.<<ETX>>

Computer vision research as a teaching tool in CS1

We have developed a computer vision teaching module consisting of materials for two or three lectures and a final project for use in a CS1 programming course. The final project is to write an image processing program with applications in computer vision. The program will read in a two dimensional array of data from a file that represents a black and white photographic image, perform one or more transformations on the data and write the transformed data to a new file. A simple image viewer program is used to display the before and after images. In addition to learning more about programming it is the intent of the project that students also have some fun with images. Most students did indeed enjoy the visual nature of the project and were surprised that they could write a program to accomplish so much after just one programming course. A few students wrote very creative transformation functions. This CS1 module is one of several developed as part of a CRCD project, sponsored by NSF. The modules are available for free use or adaptation by other instructors and institutions.

Generalization capabilities of subtle image pattern classifiers

The generalization capabilities, for learned subtle image pattern categories, of neural network and algorithmic classification techniques are described. Several neural network and algorithmic techniques have been applied to a set of feature vectors extracted from thermal infrared images which characterize the extent of whiplash injury. Thermography recently has been reported to have clinical utility in a multitude of neuromusculoskeletal disorders, particularly with soft tissue injuries such as whiplash in which there are few widely agreed upon diagnostic standards. The results of this research indicate that the backpropagation neural network produces the best classification results and provides significantly better generalization from a set of training patterns. Results of unsupervised classification of the data using clustering algorithms and the Adaptive Resonance Theory (ART2) neural network demonstrate the difficulties of learning and of generalization of patterns from such data. >

Neural network discrimination of subtle image patterns

A report is presented on a comparison between neural network and algorithmic classification techniques applied to a specific thermography program: the analysis of image patterns which characterize the extent of whiplash injury. Thermography recently has been reported to have clinical utility in a multitude of neuromusculoskeletal disorders. Of particular import is the application of thermography to soft-tissue injuries in which there are few diagnostic gold standards. Likewise, neural networks have proven to be powerful pattern separators and classifiers in a variety of real-world problems. This is primarily due to their capabilities of pattern completion and nonlinear separability in feature space. Research results show that backpropagation neural networks can accurately classify up to 90% of the whiplash thermal images, while conventional algorithmic classifiers accurately classify only up to 75%. Slight differences were found in the classification accuracy of two commercially available backpropagation implementations

Invariant feature extraction for neurocomputer analysis of biomedical images

The authors report on the application of neurocomputing techniques to a specific thermography program: the analysis of image patterns which characterize the extent of whiplash injury. A specific sequence of preprocessing steps designed to optimize neurocomputer classification of biomedical image patterns with symmetry is described. Data acquisition methods, preprocessing steps, and experimental results are discussed. The results presented are derived from incorporating neurocomputing analysis as part of an overall image characterization process. It is demonstrated that the features do represent certain characteristics of the pathology present in the images.<<ETX>>

Teaching pre-calculus students electrical engineering principles using low cost hardware

Using low cost hardware and the Visual Basic programming language, the authors have developed teaching modules which can effectively introduce basic science and technology principles at several levels from high school to freshman Electrical Engineering. They have tested these methods in a NSF sponsored course designed to teach high school teachers how to introduce science and technology to their students. To illustrate how the software tool and inexpensive hardware can be used to aid learning and teaching electrical and computer engineering concepts in the classroom, the authors present example experiments.

A plastic self-adaptive learning machine for pattern recognition

A family of neural networks and learning algorithms is introduced: the plastic self-adaptive learning machines (PSALM), together with a new interpretation of these neural networks as hyperpolyhedra in the N-dimensional Euclidean space. These networks self-adapt to a continually changing environment by properly changing the orientation of the faces of a hyperpolyhedron as well as its volume. The current structure of the hyperpolyhedron reflects the structure of the current outside world. The network optimally classifies its noise-distorted excitations into categories, after a competition between all possible categories. New categories can be created, and the old ones can be changed, or be forgotten if they are not used for a long time.<<ETX>>

Game teleporter: A development tool for everyone

The Game Teleporter is an application that translates one file format to another without any user intervention. The application itself is made up of different combinations of input plug-ins and output plug-ins. For example, the application can take an Adobe Flash file as an input and then output a Playstation Portable binary that can be run on the specified target platform and have the same functionality as the original Adobe Flash file. Flash is a file format commonly used by Web developers to create animated and interactive Websites. Using Flash as an input opens up development to people who might have great ideas for games, but cannot display them on the desired target as they do not know how to write programs for it. This tool is applicable at all levels of education. A very simple input/output plug-in combination allows elementary school children to create a game simply by drawing pictures. This can enhance their interest in learning computer programming. As shown in this paper, students used this tool with a customized graphical input plug-in to create a basic game that will run on both Microsoft Windows and on a Sony Playstation Portable. College students can also utilize this application by creating and modifying the plug-ins to adapt to whatever the current technology requires. This paper lays out how the Game Teleporter functions, and how it can be used in education at all levels to generate better computer programmers for the future.

Web Interface Development Environment (WIDE): Software Tool for Automatic Generation of Web Application Interfaces

Using web applications has become a common solution to access and manipulate information remotely. Platform and device independency can be achieved most efficiently using matured web standards. By combining available standards and web technology, complex interactive web applications can be created. However, building even basic web applications is limited to expert software developers. Even though end-users have a desire to create one, the hurdle to learn and combine multiple technologies altogether is often too high for their needs and skills. In this paper we present WIDE, a new software application designed to assists end-users develop web applications visually and thus minimize their efforts.