Recognition: Advanced Topics

Using Advanced Topics in Recognition

If you provide a grid in which the user is to write characters or words, you need to
use an

rcGridInfo

frame to define the grid to the text recognizer. For example,

the

protoCharEdit

system prototype uses an

rcGridInfo

frame internally to

define the input areas (cells) in the comb view it provides.

The recognizer uses the information in an

rcGridInfo

frame to make character-

segmentation decisions. You can use the

rcGridInfo

frame in conjunction with

an

rcBaseInfo

frame to provide more accurate recognition within boxes in a

single view. Recognition in the most recently used grid box begins as soon as the
user writes in a new box in the grid.

The NewtonScript code used to create the grid shown in Figure 10-4 looks like the
following example.

rcGridInfo := {

boxLeft: 100,// x coordinate of left of top-left box
boxRight:145,// x coordinate of right of top-left box
xSpace:55,// x distance from boxLeft to boxLeft
boxTop: 50,// y coordinate of top of top-left box
boxBottom:95,// y coordinate of bottom of top-left box
ySpace:55// y distance from boxTop to boxTop
};

To obtain the best performance and to conserve available memory, create your

rcGridInfo

frame by cloning the frame provided by the

ROM_canonicalCharGrid

constant. Store your frame in a slot named

rcGridInfo

in your view's

recConfig

frame.

For a detailed description of the

rcGridInfo

frame, see "Data Structures Used in

recConfig Frames" (page 8-24) in Newton Programmer's Reference

Creating Single-Letter Input Views

10

The following code fragment creates a single-letter input view's

recConfig

frame. This frame, which includes

rcBaseInfo

and

rcGridInfo

frames, is

based on the

ROM_rcSingleCharacterConfig

frame supplied by the system.

// specify box (or horizontal array of boxes)
// into which character(s) are written.

myView := {

recConfig: ROM_rcsinglecharacterconfig,

...}

// height of a lowercase letter
constant kSmallHeight := 11;