399 
RASTER-TO-VECTOR CONVERSION 
WRIS and VICAR. WRIS is a vector (polygon) based geographic informa 
tion system primarily designed for timber management use. It "provides 
a means of collecting, processing, storing, retrieving, updating, and 
displaying geographic data, and makes possible the performance of log 
ical operations on these data" such as merging, overlaying, and area 
measurements (Russell, et al., 1975, p.l). 
The timber strata map of the Eldorado forest produced by FOCIS is in 
raster-image format generated via VICAR/IBIS (Video Image Communication 
and Retrieval/Image Based Information System) software. VICAR 
(Castleman, 1979) is the image processing software developed at the 
Jet Propulsion Laboratory for interrogation of image data returned by 
the nation's unmanned space exploration program (e.g. Pioneer, Viking/ 
Mars, Voyager/Jupiter/ Saturn). Over the past decade, the system has 
been extensively enlarged to handle earth resources applications re 
search. IBIS (Bryant and Zobrist, 1981) is the geographic information 
system component of VICAR. 
Raster-To-Vector Conversion Algorithm. The basic algorithm was devel 
oped by D. A. Nichols (1981) and consists of three separate tasks: 
1) The first task involves identifying all raster polygons in the image. 
A raster polygon is defined as any number of hortizonally or vertically 
contiguous pixels of common label. The pixels in each raster polygon 
are assigned a new common label which is unique to each polygon, and 
the image is output with an accompanying directory for relating class 
labels bo the new unique label system; 2) The second task involves 
creating the line segments which define the polygon boundaries. Nodes 
are labeled and the two adjacent polygons identified; 3) The final task 
traverses the polygon boundary file to connect appropriate adjacent 
line segments and produce a vector file. The Nichols algorithm codes 
polygons in a clockwise manner with islands coded twice: first in a 
counter-clockwise manner and stored at the end of the host polygon's 
coordinates, and then again as a regular polygon. Lollipops (i.e. 
psuedo islands connected by one node to the host polygon) are treated 
as regular polygons. An example of the raster-to-vector conversion 
product is provided in Figure 5. 
POST-PROCESSING TECHNIQUES 
The three post-processing techniques that have been investigated can be 
grouped into two catagories based on their relationship to the clas 
sification phase of the timber stand delineation problem. In the first 
group, the modified Davis and Peet technique and 'IBIS' technique are 
both spatial filtering procedures which are applied after classifica 
tion (i.e. post-processing or post-classification). In the second 
group, the 'Labeling of Manually Delineated Polygons' technique replaces 
the classification phase entirely, and represents a psuedo raster-to- 
vector conversion technique. 
Modified Davis and Peet Approach. The spatial filtering algorithm 
adopted for FOCIS is a modified form of the one developed by Davis and 
Peet (1976). This algorithm has the advantage of removing all groups 
of pixels below a user-specified minimum size for each class in the 
image. This feature allows FOCIS to meet the ten acre minimum required 
by the Forest Service. 
The first step in spatial filtering is the removal of all single pixel 
polygons. This step is accomplished by passing a 3 x 3 moving box