Two considerations are important for interactive image spreadsheets.
First, users
want the ability to point to objects/features in the image, which
requires input. Since we cannot require the user to point to the same
object every time a cell needs to be calculated, interactive results
are cached as long as necessary. Whenever appropriate the user can
change them again. The second challenge centers around the coordinate
system. For instance, the user might load images at resolutions of 300
DPI or 600 DPI. If (s)he then would select points in a 300 DPI image,
then straightforward converting those positions to a 600dpi image will
result in wrong positions. To resolve this problem we rely on an
orthonormal coordinate system that we can alter as we go along.
Interactive User Input
Cells can have various statuses such as 'recalculating', 'should be
recalculated', 'error' and others. Two states are new compared to
standard spreadsheets. These are 'require user input' and 'user input
allowed'.
Require user input - when a formula is entered that requires user
input, such as (select A0), a formula that will ask the user to select
a position in the image, then the cell will be colored orange. In order
to enter the user input, select the cell and go to Spreadsheet|start
user input'.
User input allowed - when a cell that requires user input already
has user input it will be colored yellow. The user can then change the
input by selecting the cell and marking the cell again as 'start user
input'.
The Orthonormal Coordinate System
Because the
spreadsheet has many operations on multiple images, it
needs to know how to map a position in one image to another image. E.g;
a specific calibration spot should always be at the same virtual
location, regardless of the image. To achieve this, the spreadsheet
associates with every image a virtual coordinate system that can be
modified as necessary. This solves the problem of images that differ in
translation (a calibration spot located at different positions),
scaling (a difference in time based mass seperation might lead to
different zoom factors), image resolution and image aspect ratio.
The
orthonormal coordinate system is drawn using horizontal and vertical
lines through (-1,-1), (0,0) and (1,1). On every 0.1 step between -1
and 1 a dotted line is drawn.