Published references

• Edge and Corner Detection by Photometric Quasi-Invariants
  Discusses classification of illumination features given basic lighting information
• Tensor Based Feature Detection for Color Images
  A tensor based system for color image feature detection.
• principles for automatic scale selection
  A paper that explores the effects of scale in feature detection.
• SUSAN - A New Approach to Low Level Image Processing
  Describes the SUSAN algorithm, a non-derivative based method for robust feature detection.
• DUAL PHOTOGRAPHY
  Not necessarily about computer vision. This paper discusses the 'Dual Image' given a known lightsource and camera location. May be interesting in terms of photometric invariant features.
• Correspondence algorithms
  An exhaustive taxonomy of correspondance algorithms in Stereo vision. Many references, over 20 algorithms. An interesting test-bed for benchmarking new works.
• Detecting, localizing and grouping repeated scene elements from an image.
  This paper discusses a method using corellated windows to find repeated elements (textures!) in imagery. it attempts to be robust against local changes, by assigning afine transformations to each corellated window. it seems like a pretty good way to extract texture and feature boundaries, but its results left a fair bit to be desired
• Shape Descriptors for Maximally Stable Extremal Regions
  This paper discussed a blob detector, Maximally Stable Extremal Regions algorithm. these work by analyzing windows (over the intensity space) of the image, and grouping locally related. (that is, features that are stable as the window is moved over the intensity space are recognized). this algorithm is in heavy use in stereo vision corellation.
• Maximally Stable Colour Regions for Recognition and Matching
  This is an extension to color imagery from the last paper, with some work on clustering for beter back-end recognition.
• RobustWide Baseline Stereo from Maximally Stable Extremal Regions
  The first work on this algorithm, being MSER. I don't really understand certain aspects of this algorithm.
• Multiframe Image Point Matching and 3D Surface Reconstruction
  IEEE Transactions on pattern analysis and machine intelligence, vol. PAM1-5 No. 2, March 1983
  paper uses an 8 Camera configuration to match distances to points that have been corellated on a surface.jmm (joint moment method) and mwv(moving window variance) approaches are used to correlate (via epipolar geometry) at high speed from 8 carefully placed co planar cameras. Contains good discussion of correspondence problem difficulties, but attempts are overcomplicated.
• Model-Based Three-Dimensional Interpretations of Two-Dimensional Images
  IEEE Transactions on pattern analysis and machine intelligence, Vol PAM1-5, No. 2, March 1983
  presents the ACRONYM system, which takes a vectorized model to compare against 2D images in an attempt to extract matching image features. paper does not do a great job on covering issues like false positives, or performance under more general circumstances.
  
• The Stanford Cart and the CMU Rover
  Proceedings of the IEEE, Vol. 71, No. 7, July 1983
  An early use of vision system in robotics control. Rover uses a camera mounted on a translational stage to capture and digitize images, which are then analyzed by an intrerest opperator. Computational power was very limited, but this experiment gives insight into what factors may be more important than others. accuracy and cleaness of image for example, were not needed for the system to function.
• Object Tracking: A Survey
  ACM Computing Surveys, Vol 38, No 13, 2006
  Broad paper discussing fundamentals and categories of different feature tracking techniques. Lots of good Ideas, and relativly thourough analysis of most topics covered. Some topics include: feature detection, feature descriptors, and applied classification algorithms (SVM, etc).
• Basic Similarity Concepts
  Discusses the problem of similarity, compares common methods (i.e. distance), discusses why there is a need for heirarichal extensibility in these measurements.
• Scale Space Theory in Computer Vision
  I haven't read this book, but I see this guy's work everywhere. contains a link to a thesis on the same topic. Discusses the fundamental effects that scale has on computer vision applications.
• Computer Vision
  An older, very interesting looking book on computer vision and EVERYTHING related, both practically and theoretically. Entire book may be downloaded free of charge. lots of interesting material to work through in here, including volumetric representaions and much more.

Good background material

• Epipolar visualization
• Epipolar Math and equations
• Various depth cues in one and two eye systems
• Retina Diagrams
• Ganglion cells

Practical Resources

• Logitech Webcams
  Logitech makes many different kinds of webcams, all high quality and a large price range. supposedly very easy to use from a development point of view. This site has several examples.
• Microsoft Robotics Studio
  Integrated robotics (including vision) development environment with loads of support for simulation, visualization, and control.
• DirectShow
  The Windows API for getting at raw frames of a webcam. Fast, but appearantly complicated API.
• Logitech Webcams under Linux
  a friendly way to develop webcam software under linux.