VIZA 654 / CSCE 646
DIGITAL IMAGE
Synopsis

This course provide a thorough grounding in the state of the art in the treatment of digital images, particularly within the context of computer graphics, and digital effects. The course is designed to prepare students to understand existing systems for storage, display, transformation and manipulation of digital images write their own software for working with digital images undertake creative work and research involving digital images. We cover tools and techniques for the generation, handling and analysis of two-dimensional digital images. Image representation and storage, display, media conversion, painting and drawing, warping, color space operations, enhancement, filtering, and manipulation. Just as digital sound has become the standard for high-quality audio recording, the digital image is becoming the standard form of electronic image. Digital images have the advantages of lossless storage, transmission, and retrieval. Their form greatly facilitates generation, manipulation, and display within a computing environment, and they provide a natural syntax for image representation that pervades the world of computer graphics and visualization. Thus, an understanding of the nature, form, and technology of the digital image is essential to a visualization practitioner. Upon completion of this course, students will know the state of the art in the treatment of digital images in the context of computer graphics, and digital effects. They will understand existing systems for storage, display, transformation and manipulation of digital images, they will be able to write their own software for working with digital images and they will be able to undertake creative work and research involving digital images.



EXAMPLES OF TOPICS
FILTERS
Implementation of Nonstationary Filters
In this project, students develop non-stationary filters that can be applied to an image file to produce unusual images. The goal of this project is to control filter kernels for each pixel to obtain aestetic results that cannot be directly created by using any stationary-filters, which use static kernels, that are provided by existing image processing software. In this example, Carl Van Huyck manipulated an image by using it both the input and the control for how the filter is applied, changing both the size and composition of the matrix.
Implementation of Nonlinear Filters
In this project, students develop non-linear filters that can be applied to an image file to produce unusual images. The goal of this project is to use non-linear operations that still use (stationary or non-stationary) kernels. Since the operations are not linear, the results may not necessarily be provided by existing image processing software. In this example, Cameron Black manipulated an image using a dilation operation and an non-stationary kernel. Unusual and changing blocks of colors in the final image are caused by both shapes and non-stationary nature of kernels.
DITHERING & SCREENING
Implementation of Dot Diffusion
In this project, students implement a variety of dithering methods such as Floyd-Steinberg error diffusion, ordered dither or Artistic Screening. In this particular example, Zhen Li implemented a version of Donald Knuth's Dot Diffusion algorithm. Note that how the result resembles to the images that are obtained by physical screening process.
FINAL PROJECTS
Development of New Methods
In this project, students develop their own methods to create interesting imagery. In this case, Sebastian Kawar created an animation generating program that tracks mouse movements and changes images using color and brightness of pixels. The left image is a frame from the animation shown right.
Implementation of Existing Work
In this project, students implement a version of a recently published work. In this case, an input image is altered by assigning various sections of the image random rotation and translation values. Emily Harris got the idea for the algorithm from Reynold Bailey and Cindy Grimm's paper, "Creating the Illusion of Motion in 2D Images", in which they attempt to give standard images the appearance of having motion, as in many paintings that use spatial imprecision. Wan Gogh portrait is a triangulization of an original image using Hugues Hoppe's Progressive Meshes, by Zhen Li.