Additional Resources and Links

AI Magazine Volume 20 Number 2
http://www.aaai.org/Library/Magazine/Vol20/20-02/vol20-02.html
Note: Click on the link "Computers Seeing People" to pull up an abstract of the article, and then click on the title link in the new window to open up the pdf of the actual article.
This is the cover article from AI magazine's summer 1999 issue, and provides perhaps the single best overview of the emerging field of computer vision and its applications to human/computer interaction. It is concise and understandable, and discusses why computer vision is needed, what has been done in the field so far, and where we need to go. The author argues that for computers to go beyond their current role as machines that just "sit in the corner of the room," they need to be equipped with vision systems that allow them to see and interact with humans. The article goes over numerous topics, such as people tracking, face and gesture recognition, etc.
Rating:1

Al Qual Summary: Vision
http://www.cs.dartmouth.edu/%7Ebrd/Teaching/AI/Lectures/Summaries/vision.html
Presents an in-depth discussion of specific techniques and algorithms involved in vision recognition. The material presented is very useful and provides a good introduction to the technical aspects of vision. This is the kind of information that one would expect to find in a college textbook on the topic. The technical discussion is somewhat difficult to understand and is not suited for everyone.
Rating:2

Beautiful Vision
http://redherring.com/mag/issue81/mag-beautiful-81.html
This is an article from Red Herring. It provides a general discussion of how vision has been and will be extremely important in the process of making robots ever more advanced. Seeing as how this is a business magazine, the large part of the discussion is dedicated to how vision may be used to allow robots to more intelligently perform industrial tasks and cut down costs. With the ability to see, robots will become more dynamic and robust in their ability to perform different tasks. The article is very readable.
Rating: 3

The British Machine Vision Association
http://www.bmva.ac.uk/
This is the website of the British machine vision association. It contains good but general information about the field of machine vision. It discusses the objectives and current activities of the association, as well as applications of machine vision. For example, the website gives a brief overview of how machine vision may be used for quality control, surveillance, observing the environment, entertainment, and medicine.
Rating: 3.5

Computational Vision
http://www.vision.caltech.edu/
This is a link to Caltech's vision department. It provides an overview of what research is being done there and who it is being done by. It has links to numerous papers and publications in vision research, but these papers weren't really intended for the average person to read. There are also links to other related websites, but even these are similar in that they are more tuned towards the minds of academia. In the end the average person will come off with a better general understanding of some of what is going on in vision research, but from this site alone they probably will not understand the specifics of how things are being done.
Rating: 3

The Computer Vision Homepage
http://www-2.cs.cmu.edu/afs/cs/project/cil/ftp/html/vision.html
A well-maintained site with hundreds of links to everything the average person could possibly want to know to know about vision recognition. The links are conveniently categorized to make it easier to find useful information. However, the website itself does not have a substantial amount of original material, and the sheer volume of links limits its usability. This is an in-depth site can in no way be considered a "concise" rundown of vision recognition.
Rating:3

CVonline - Compendium of Computer Vision
http://www.dai.ed.ac.uk/CVonline/
Called an "online compendium," this is a website that goes over many specific aspects of computer vision, mostly from a computer science perspective. It does not provide a general overview, but instead it has a hierarchical listing of links to hundreds of specific topics in the field. Its main use is in allowing users to find a specific "techniques or ideas" in computer vision. For example, the site has five links dedicated to how computer vision can be used in agricultural crop treatment and control. These are very specific solutions to very specific problems. The readability depends on the specific link visited, but generally the links are to research papers and specific topics, and these papers are somewhat esoteric.
Rating: 2

Document Image Understanding
http://www.cedar.buffalo.edu/Publications/TechReps/Survey/survey.html#TEXT
Character Recognition, also known as Optical Character Recognition or OCR, is concerned with the automatic conversion of scanned and digitized images of characters in running text into their corresponding symbolic forms. The ability of humans to read poor quality machine print as well as text with unusual fonts and handwriting is far from matched by today's machines. Most errors were caused by one of three image quality deficiencies: underlines in text, "I" and "i" confusions, and distortion of "e"'s. Underlines occurred in 27 words and descenders in these words were often completely obscured. This led to recognition failures in all three packages. The dots on "i"s were often smeared together with the body of the "i" thus causing the case confusions. For "e", the small opening of the "e" was often closed due to the low quality of the images. In sum, recognition deficiencies seem primarily due to imprecise character definitions. The document is quite easily read by a human unfamiliar with the domain. Word models can be represented in different ways. An obvious method is to store the list of legal words, or lexicon. Other methods include n-grams (legal letter combinations), Markov models - which capture first and higher order letter transitional probabilities - and n-gram probabilities. Three distinct approaches to utilizing word models in character recognition can be identified: character-based word recognition, segmentation-based word recognition and word shape recognition.
Rating: 1

ETHOS News: Vision Recognition Systems for the Future
http://www.ethoseurope.org/ethos/lit3.nsf/0/4009ecc027f5e1f18025652b002f078c?OpenDocument
Workers at Microsoft's Research Vision Technology Group are aiming to develop PCs that will recognize your face, understand your gestures and can be turned off by a simple shake of the head. While many PCs offer speech recognition and speech synthesis systems, they cannot "see" their users or recognize their speech. Videoconferencing systems will send still or moving images of a PC user down telephone lines to another PC user, but the PC simply displays the video image, it does not recognize the users or understand what they are doing. Microsoft has therefore been developing vision recognition technology and in a recent demonstration, a PC was programmed to recognize a senior researcher from their laboratory when he was present. The PC was equipped with a camera on top, which captured his image and displayed it inside a rectangular box or "vision zone" on the PC screen. The vision zone is used by the PC to detect whether anyone is present. When the researcher moved into the vision zone, the PC automatically displayed his work on the screen. It is able to do this due to sophisticated vision processing software, which could be refined to recognize individual users and deny access to unauthorized users, according to Microsoft. Another example of the system's capability was when the researcher played a game of checkers by simply moving his hands in front of the PC. In the long term, Microsoft is aiming for a system that can recognize eye movements, so that software could be controlled by eye movement. The playing of instruments on a PC using gestures is also a possibility, where, for example, a hand clap could create a cymbal sound. The people who would benefit most from vision recognition technology are those with disabilities. For example, a speech recognition system could be enhanced by a vision recognition system so that PCs could lip-read. However, millions of other people who find it difficult to operate a computer keyboard or mouse could also benefit. While vision recognition technology is some six or seven years away from reaching the mass market, Microsoft say that more and more PCs will offer some form of it, and believe that within three years, the majority of computers will have a built-in camera.
Rating: 3

GRACE: The Social Robot
http://www.palantir.swarthmore.edu/GRACE/
This website was created by the some of the developers of Graduate Robot Attending a ConferencE. It contains a general discussion of GRACE the social robot's design and activities. GRACE was a joint effort between a number of institutions, and is a robot designed to attempt to fulfill the requirements of the Robot Competition. In this competition a robot must, quite simply, attend the yearly national conference on artificial intelligence. GRACE does not have any arms or legs, and simply rolls along to get around. In July 2002, GRACE the robot, with some help from a human assistant, successfully navigated her way through a lobby, up through and elevator, and gave a lecture about herself to a group of hundreds of researchers. The website presents only one implementation of vision recognition, and is quite general in its discussion.
Rating: 3

Howstuffworks "How Augmented Reality Will Work"
http://computer.howstuffworks.com/augmented-reality1.htm
The basic idea of augmented reality is to superimpose graphics, audio and other sense enhancements over a real-world environment in real-time. Sounds pretty simple. Besides, haven't television networks been doing that with graphics for decades? Well, sure -- but all television networks do is display a static graphic that does not adjust with camera movement. Augmented reality is far more advanced than any technology you've seen in television broadcasts, although early versions of augmented reality are starting to appear in televised races and football games, such as Racef/x and the super-imposed first down line, both created by SporTVision. These systems display graphics for only one point of view. Next-generation augmented-reality systems will display graphics for each viewer's perspective. Augmented reality is still in an early stage of research and development at various universities and high-tech companies. Eventually, possibly by the end of this decade, we will see the first mass-marketed augmented-reality system, which one researcher calls "the Walkman of the 21st century." What augmented reality attempts to do is not only superimpose graphics over a real environment in real-time, but also change those graphics to accommodate a user's head- and eye- movements, so that the graphics always fit the perspective.
Rating: 1

The Industrial Physicist
http://www.aip.org/tip/INPHFA/vol-9/iss-1/p18.html
A webpage that discusses recent developments in vision recognition. Topics discussed include face recognition, augmented vision systems, and computerized video surveillance. Recent advances allow computers to isolate objects and track them in a moving image. One computer system can track a human face as it moves around a room. Another system, used for surveillance, allows traffic cameras to follow cars and other objects that are present in their feeds.
Rating:2

Penn State Computer Vision Lab
http://vision.cse.psu.edu/
The Penn State Computer Science Lab Vision page presents a review of the group's current projects. It has a brief overview of vision recognition and its applications, and has information about some of the current projects going on there. According to the website, current research falls into the following three categories: video content analysis, augmented reality systems, and gesture recognition. Video content analysis basically means looking at the contents of an image and making interpretations. Augmented reality systems involve computers superimposing virtual images over a human's actual view of the real world. Gesture recognition will allow humans to more easily communicate with computers, and involves using computer vision to interpret human gestures.
Rating: 3

Seeing is Believing: Computer Vision and Artificial Intelligence
http://www.acm.org/crossroads/xrds3-1/vision.html
An excellent article that gives a concise overview of using computer vision to interpret digital images. This article is packed with information and is in an easy-to-read format. It goes over different techniques of having agents use knowledge to analyze digital images, such as probability theory and Bayesian networks.
Rating:2

UC Berkeley Computer Vision Group
http://www.cs.berkeley.edu/projects/vision/vision_group.html
This is Berkeley's vision web site, and contains information on their computer vision projects. Unlike other university websites, however, there is more here than just hard-to-understand research papers and publications. Publications are categorized into different research topics, and overviews of the different topics give the reader a better overall understanding of the research being done. Research categories include image and texture analysis, viewing "humans and their activities," and computer surveillance techniques.
Rating: 2

Using Robotics to Scare Birds
http://robots.net/article/215.html
A story from Discover News reports that researchers at the LSU Aquaculture Research Station have developed a floating, autonomous robot platform powered by solar panels. The floating robot searches for birds using a vision recognition system and scares them away by bumping into them or squirting them with a water gun. The robot floats in catfish ponds to protect the stock which is frequently depleted by hungry birds. They hope the robot will be able to replace the dangerous poisons and annoying loudspeakers currently used to stop bird predation at catfish ponds.
Rating: 4