Hyunmo Kang
Zoomable User Interface
CMSC 838B
March 3, 1999
 

Image-Browser Taxonomy and Guidelines for Designers

Catherine Plaisant, David Carr, and Ben Shneiderman.
University of Maryland

Summary :

The purpose of this paper is to give useful guidelines to the 2D Image-Browser designers by showing the Image-Browser Taxonomy in various aspects. So here the author introduces an informal specification techniques to describe 2D browsers and a task taxonomy, suggests design features and guidelines, and assess existing strategies (but only focussing on the tools to explore the selected image).
This paper is roughly composed of  five big sections; Browser Specification(about DMsketch), Review of some classic techniques used in existing 2D browsers and their variations, Task Taxonomy, Browser Taxonomy for presentation aspects, and finally Browser Taxonomy for operation aspects.

1. DMsketch(direct manipulation sketch) is a method developed to help designers exchange and record ideas more quickly and clearly than a formal specification language. It consists of several primitives(movement constraint, proportional size constraint, field of view(six variations), fitted projection), composite objects(which are used to simplify the specification), and commands.
2. Multitude of browsers : some classic techniques and their variations in existing systems can be classified as follows.
       - Detail only browser : the most common technique.
       - Single window with zoom and replace : the user marks a rectangular area which is magnified and replaces the original image.
       - Single coordinated pair (overview-detail) : combining displays of the overview and a local magnified view.
       - Tiled multilevel browser : combining global, intermediate, and detailed views.
       - Free zoom and multiple overlap : users are free to specify, move, reshape and delete every window as they wish.
       - Bifocal view browser : using a magnifying glass metaphor.
       - Fish-eye view : distorting the magnified image so that the center of interest is displayed at high magnification, and rest of the image is compressed.
3. Task Taxonomy : five classes of tasks users accomplish with image browsers.
       - Image generation : an overview is important, but most of the time is spent at a detail level.
       - Open-ended exploration : navigation must be fast.
       - Diagnostic
       - Navigation : zooming and panning occur only occasionally.
       - Monitoring : window management is an important issue.  
4. Browser Taxonomy for presentation aspects
       - Static presentation :
           * Single view browsers : Detail-only, Zoom-and-replace, Fish-eye
           * Multiple-view browsers : three important considerations when designing multiple view browsers are
                    Window-placement strategy (SSROD, view layout), Coordination, Global view.
       - Dynamic aspects :
           * Quality of the update(continuous, jump, updates a preview then add detail when stops)
           * Nature of the update(expansion, explosion, distortion)
           * Zooming factor(small/medium/large)
5. Browser Taxonomy for operation aspects
       - Manual :
            * Zooming(zooming location specification, zooming factor, destination window, zooming out)
           * Panning(scroll, sticky hand, arrow keys)
       - Automation :
            * Save points : speeding navigation and diagnostics.
           * Navigation 
           * Window management automation
           * Image search 

And as a conclusion, author addresses the fact that the goal of the whole complexity - caused by many options, features and parameters described in this paper- of image browser interfaces is to design the simplest and the most convenient tools that fit the task. And he also concentrates that because of the complexity of two-dimensional browsing, more careful analysis, design, and evaluation might lead to significant improvements.
 

Comments about contributions :

1. The paper is well organized and gives designers really good guidelines for designing 2D Image-Browser.
2. The paper shows a number of good examples concerning the image browsers and it is very easy to understand the description of browser features, due to the usability of DMsketch.
3. The paper indicates the problems of existing 2D-image browsers appropriately and also suggests some plausible solutions.
4. By showing not only the browser taxonomy but the task taxonomy, the paper gives designers intuition that it is very hard to implement an desirable image browser without considering the task it performs.
5. The paper performs classification of image browsers in several view points which are not mutually exclusive, such as static and dynamic representation in presentation aspects, and manual and automated operations in operation aspects, which help designers to consider the browsers to be implemented in different points of view.  
   

Questions :

1. Can this taxonomy be applied to general browsers besides the 2D image browsers ? If not, how can it be converted to fit general browsers ?
2. Do you think it is possible to describe all the existing and developing image browsers only with DMsketch ? If not, what more types of primitives are needed ?
3. What more features and techniques are needed to browse a series of images or to browse large-image databases ?   
4. Do you think it is useful to use new techniques such as semantic zooming in browsing image ? Which classification can this technique be included ?