Colin Ware, Marlon Lewis
Colin Ware
Faculty of Computer Science
University of New Brunswick
P.O. Box 4400, Fredericton
New Brunswick, Canada E3B 5A3
Email: cware@UNB.ca
Marlon Lewis
SatLantic Inc.
Richmond Terminal, Pier 9
3295 Barrington Street
Halifax, NS. B3K 5X8
1) Fisheye view - In this case we enlarge one or more
points of interest and compress the rest of the image
proportionately with distance from the point of interest.
This appears to work well for 2D information networks
but seems poor for images and maps where the distortion
is likely to result in a confusion [2].
2) Magnifying glass - problem is a loss of surrounding
context (e.g [1]).
3) Separate magnified window with place holder in the
base window to show the source. In the present paper we
describe two solutions based on this option but with a
number of additional features
The requirement of the prototypes described here was to
allow an operator to rapidly interpret a large image of up
to 12,500x12,500 using a computer interface. The
particular target application was synthetic aperture radar
imagery, but the problem is general to most high
resolution images displayed on a screen. The two
DragMag Image magnifier prototypes described in this
paper were designed to solve the problem of continuous
smooth zoom without loss of context and to allow for both
coarse and fine translation movements. Two different
designs were implemented as prototypes and evaluated.
FIGURE 1
DragMag Prototype 1
Control of the position of the magnified region and the
degree of magnification is by three methods
All of the above adjustments result in real-time changes in
the Magnified image and in the position and size of the
Mag Window and connecting lines. In the prototype the
Zoom Windows cannot be repositioned.
FIGURE 2
DragMag Prototype 2
Control of the position of the magnified region and the
degree of magnification window is by three methods.
To create a DragMag, the user positions the cursor on the
base image at the region of interest. A DragMag instantly
appears enlarging the selected center of interest. If
desired another region can be selected and a second
DragMag will appear. In the prototype implementation up
to three separate DragMags are allowed.
Part of the idea behind this interface is that the base image
should be as large as possible and should ideally cover the
entire screen in order to permit the best possible overview.
The two DragMag prototypes also have different
individual strengths. The degree of magnification in Drag
Mag 1 is controlled with a slider. This will normally
require a mouse movement combined with an adjustment
movement. While this is acceptable it was felt that some
quicker method should be found for some applications.
The fact that the windows were non-overlapping and in
fixed locations was both an advantage and a disadvantage.
It was an advantage in that it is simple, but the base
window is necessarily small and the number of zoom
windows must be fixed. However, the user is freed from
the burden of window positioning.
Magnification in DragMag 2 is done by distance between
the window to the point of interest. This works well but
is judged to be somewhat less useful for systematically
scanning the data for features that are invisible in the base
image. However it is judged to be better for rapidly
magnifying features that have already been identified in
the base image. Thus it is probably better for medical
applications, such as the examination of detailed X-ray
images. The fact that the entire screen can be used for the
base image is good but places a burden on the user who
has to consider a good placement for windows.
We recommend that the basic design given in Prototype 1
be implemented but with some features from Prototype 2.
It should be possible to have a base window the size of the
entire screen so as to allow the best possible understanding
of the entire image. The DragMag Zoom windows should
be sizable and placable, but otherwise function as in
Prototype 1. The Zoom Windows should not be allowed
to overlap and should each be confined to one of the four
screen quadrants to simplify placement. This will allow
for between one and four DragMags to be placed at any
one time.
It should be possible to mark and label interesting features
in the Base window and in the Zoom Window. This will
allow for the system to provide high quality illustrations to
document things that have been identified.
Abstract
Computer image magnification often results in the loss of
the contextual information. Two image magnification
prototypes are presented that solve this problem. In both
the base image is shown at all times with lines connecting
the magnified region on the base image to the magnified
image. Both prototypes allow for smooth pan, scroll, and
zoom of large 2D images with both coarse and fine
translation movements. The second design implements a
magnifying glass metaphor whereby the degree of
magnification is controlled by the distance of the enlarged
image to the magnified region of the base image.
Keywords:
Data display, Pan, Scroll, Zoom.
Introduction
When panning and zooming through a large image the
biggest problem is to avoid getting lost. At high
magnifications all context is lost and it is usually
necessary to zoom out, before zooming in to another point
of interest. One of the best solutions to this problem is to
maintain a plan view of the entire scene at all times, and to
show magnifying windows. However there are various
ways in which this might be implemented.
DragMag Prototype 1
The first design is illustrated below and in the CHI-
VIDEO review videotape. There basic elements are: a
Base Image which consists of a view of the entire image
filtered down so that it fits inside a window, the Mag
Window , a region of the base image that is being
magnified (highlighted with a color tinted patch,) and the
Zoom Window, which is the magnified region. Lines
connect the Zoom Window to its associated Mag Window
to show the source of the information without ambiguity.
Lines connect the corners of the Mag Window to the
Zoom Window.
DragMag Prototype 2
The basic elements of DragMag 2 are the same as
DragMag 1 as can be seen in this diagra
EVALUATION OF THE PROTOTYPES
Both Drag Mag prototypes are successful in allowing the
user to scan an image with large coarse movements (by
moving the Mag Window) or with fine control (by
dragging in the Zoom Window). This aspect of the
interface is highly successful, it is natural and has a direct
manipulation "feel". The idea of attaching the Mag
Window to the Zoom Window works well in maintaining
a visual cue as to the source of the information.
