RECOGNIZING CONVEX BLOBS
by
J . Sklansky
University of California
Irvine, California
92664
Abstract
we develop
blobs.
In
t i o n o f the
intuitive.
Because o f t h e d i s c r e t e n a t u r e o f t h e memory
and l o g i c o f a d i g i t a l c o m p u t e r , a d i g i t a l com­
p u t e r " s e e s " p i c t u r e s i n c e l l u l a r f o r m , each c e l l
c o n t a i n i n g a number t h a t r e p r e s e n t s t h e
d e n s i t y o f the viewed o b j e c t a t t h a t c e l l .
In
p a r t i c u l a r , when t h e p i c t u r e i s b i n a r y , e a c h c e l l
h o l d s a 1 or 0, depending on whether or n o t t h e
viewed object is p r o j e c t e d onto that c e l l .
The
convexity of c e l l u l a r blobs - i . e . , binary singly
c o n n e c t e d c e l l u l a r f i g u r e s - i s d i s c u s s e d and
d e f i n e d in terms of the continuous blobs of which
the c e l l u l a r blobs are images.
the elements of a theory of convex
the I n t e r e s t o f b r e v i t y , the p r e s e n t a ­
theory i s p a r t l y nonrigorous and
For r i g o r o u s p r o o f s , see Reference 4.
Statement of the Problem
A f i g u r e i s d e f i n e d t o b e convex i f i t con­
t a i n s the l i n e segment t h a t j o i n s any two p o i n t s
o f the f i g u r e .
Otherwise the f i g u r e i s concave.
A t h e o r y o f convex c e l l u l a r b l o b s i s s k e t c h e d ,
and t h e u s e o f t h e " m i n i m u m - p e r i m e t e r p o l y g o n " i n
an a l g o r i t h m f o r t e s t i n g the convexity of c e l l u l a r
blobs is described.
Introduction
I n d e s i g n i n g o r programming d i g i t a l machines
to recognize two-dimensional connected o b j e c t s ,
one i s o f t e n c o n c e r n e d w i t h t h e g e o m e t r i c p r o p e r ­
t i e s of the presented o b j e c t s .
Examples o f s u c h
p r o p e r t i e s are convexity, elongatedness, t h r e e lobedness, e t c .
Consider the c e l l u l a r blobs i l l u s t r a t e d i n
Figure 2.
I n t u i t i o n t e l l s us t h a t Blob A is a
c e l l u l a r image of a convex o b j e c t , and t h a t Blob B
is a c e l l u l a r image of a concave o b j e c t .
Blob A,
considered as a continuous f i g u r e , is c l e a r l y
concave, as shown by the d o t t e d l i n e . Hence we
need t o f i n d a r e a s o n a b l e , i n t u i t i v e l y s a t i s f y i n g
d e f i n i t i o n o f "convex c e l l u l a r b l o b . " The proper­
t i e s we b e l i e v e such a d e f i n i t i o n must have are
discussed in the next two paragraphs.
We t h i n k of " c o n v e x i t y " as a form of "smooth­
n e s s . " I . e . , the more convex an o b j e c t i s , the
smoother it i s . When we ask whether a c e l l u l a r
image J is convex or concave, we are t h e r e f o r e ask­
i n g whether the smoothest o b j e c t q, such t h a t
I ( q ) ■ J, is convex or concave, where
l a r image of q.
Thus if we can f i n d any plane
f i g u r e , say r, such t h a t I ( r ) - J and such t h a t r
i s convex, then a l l o b j e c t s smoother than r , say
such t h a t
w i l l a l s o b e convex.
The p r o p e r t i e s o f c o n t i n u o u s c o n v e x f i g u r e s
a r e w e l l d e f i n e d and u n d e r s t o o d 1 .
But a computer
" s e e s " these o b j e c t s in the form of c e l l u l a r
r a t h e r t h a n c o n t i n u o u s i m a g e s , each c e l l h o l d i n g
a number t h a t r e p r e s e n t s t h e o b j e c t ' s p r o j e c t i o n
into that c e l l .
Hence i t i s i m p o r t a n t a ) t o d e f i n e r i g o r o u s l y
the geometric properties of c e l l u l a r blobs in
terms of the continuous o b j e c t s of which the
c e l l u l a r b l o b s a r e i m a g e s , and b ) t o d e v e l o p a l ­
gorithms that t e s t c e l l u l a r blobs f o r these
properties.
I n t h i s paper w e r e s t r i c t our a t t e n t i o n t o
two-dimensional binary objects or "blobs", i . e . ,
b l a c k f i g u r e s o n a w h i t e b a c k g r o u n d , and t o
b i n a r y c e l l u l a r "images" o f these o b j e c t s , i . e . ,
l ' s on a background of 0 ' s .
A cell holding a 1
r e p r e s e n t s a nonempty p r o j e c t i o n o f t h e o b j e c t
i n t o the c e l l .
Two e x a m p l e s o f c o n t i n u o u s b i n a r y
o b j e c t s a r e shown i n F i g u r e 1 .
F i g u r e 2 shows
how t h e s e o b j e c t s a r e u s u a l l y s e e n b y a d i g i t a l
computer.
I n t h i s f i g u r e t h e c e l l u l a r images a r e
arranged on a r e c t a n g u l a r mosaic.
Other mosaics,
such as h e x a g o n a l or i r r e g u l a r m o s a i c s , a r e a l s o
possible.
W e d e s c r i b e t h e p r o b l e m o f d e f i n i n g and t e s t ­
i n g c o n v e x i t y o f bounded c e l l u l a r b l o b s , and w e
present a s o l u t i o n .
In p r e s e n t i n g our s o l u t i o n ,
This leads us to the f o l l o w i n g p r e l i m i n a r y
d e f i n i t i o n of c e l l u l a r convexity:
A c e l l u l a r blob
i s convex i f and only i f t h e r e e x i s t s a t l e a s t one
convex f i g u r e r of which the g i v e n c e l l u l a r blob
is an image.
Searching f o r such an r is not a p r a c t i c a l
t e s t f o r c o n v e x i t y , however, because even a f t e r an
i n d e f i n i t e l y long u n s u c c e s s f u l search such an r may
s t i l l e x i s t . What we need is an a l g o r i t h m f o r
c o n s t r u c t i n g an o b j e c t p, such t h a t I ( p ) ■ J, and
such t h a t if p is concave then every other o b j e c t
whose image is J w i l l n e c e s s a r i l y be concave, t o o .
We show in Theorems 1 to 3 t h a t the "minimump e r i m e t e r p o l y g o n " answers t h i s need.
Unger's a l g o r i t h m s f o r d e t e c t i n g " v e r t i c a l con­
c a v i t y " and " h o r i z o n t a l c o n c a v i t y " are the c l o s e s t
known e a r l i e r approaches to the d e t e c t i o n of convex
cellular b l o b s .
I t i s easy, however, t o draw a
concave b l o b t h a t i s v e r t i c a l l y convex and h o r i ­
z o n t a l l y convex.
Such a blob is shown in F i g . 3.
The d o t t e d l i n e shows t h a t t h i s blob is concave.
-107-
Elements of the Theory of C e l l u l a r Blobs
Elementary Concepts of Plane Figures
A
simple curve is d e f i n e d i n t u i t i v e l y as the
curve o b t a i n e d from the continuous motion of a
p o i n t on a p l a n e , such t h a t the path of the p o i n t
never crosses or becomes tangent to i t s e l f , except
p o s s i b l y when the path r e e n t e r s i t s e l f .
A simple
c l o s e d curve is a simple curve which r e e n t e r s its e l f . A simple curve may be bounded or unbounded
at either of i t s "ends."
(For r i g o r o u s d e f i n i ­
t i o n s o f these e n t i t i e s , see A l e x a n d r o v 1 . )
If
the d i s t a n c e of p r e c i s e l y one of a simple c u r v e ' s
ends from the p l a n e ' s o r i g i n i s i n f i n i t e , the
curve i s s i n g l y unbounded; i f both o f a simple
c u r v e ' s ends are i n f i n i t e l y d i s t a n t from the
o r i g i n , the curve i s doubly unbounded.
A plane f i g u r e , o r simply a f i g u r e , i s d e f i n ­
ed here as a s e t of p o i n t s f having the f o l l o w i n g
properties.
1.
f l i e s i n a plane
2.
f
3.
f c o n t a i n s a simple curve c which is
e i t h e r closed or doubly unbounded
f c o n t a i n s the i n t e r i o r of c
f c o n t a i n s no p o i n t of the e x t e r i o r
of c
4.
5.
=
, where
is t h e empty s e t
Curve c is the boundary of f.
We u s u a l l y r e p r e s e n t a f i g u r e by a lower case
c h a r a c t e r , such as p, q, r.
A f i g u r e i s bounded i f i t l i e s e n t i r e l y w i t h ­
i n some c i r c l e o f f i n i t e d i a m e t e r .
Thus q u a d r i ­
l a t e r a l s and e l l i p s e s are bounded f i g u r e s . A
b l o b is any bounded f i g u r e .
Note t h a t if a
f i g u r e i s bounded, i t s boundary must b e c l o s e d .
A s e t o f p o i n t s s i s connected i f i t i s nonempty and i f every p a i r o f p o i n t s i n s i s c o n t a i n ­
ed in a simple curve b e l o n g i n g e n t i r e l y to s. A
s e t o f p o i n t s i s simply connected i f i t i s connect­
ed and if t h e r e e x i s t s no f i g u r e f whose boundary
l i e s i n s , b u t some p o i n t i n f does not l i e i n s .
Note t h a t every f i g u r e , a s w e have d e f i n e d i t , i s
simply connected,
A polygon is a f i g u r e whose boundary c o n t a i n s
o n l y s t r a i g h t l i n e segments.
Thus, i n t h i s paper,
a r e c t a n g l e is a p o l y g o n , but a q u a d r i l a t e r i a l
w i t h a p a i r o f i n t e r s e c t i n g o p p o s i t e sides i s n o t .
The v e r t e x angle of a polygon is the i n t e r i o r
angle between two adjacent edges of the p o l y g o n .
Note t h a t a v e r t e x angle l i e s in one of the open
intervals (0,TT),
(TT, 2TT) .
As a consequence of the d e f i n i t i o n of con­
v e x i t y , a polygon i s convex i f and o n l y i f each
of i t s v e r t e x angles is l e s s than TT r a d i a n s .
Hence every t r i a n g l e is convex.
The above obser­
v a t i o n s lead t o the f o l l o w i n g d e f i n i t i o n s . A
v e r t e x o f a polygon i s a convex v e r t e x i f i t s v e r ­
t e x angle i s l e s s than I T r a d i a n s ; i t i s a concave
v e r t e x if i t s v e r t e x angle exceeds TT r a d i a n s .
A c e l l u l a r mosaic* is a set of bounded convex
f i g u r e s ( c } , c a l l e d c e l l s , such t h a t
either
o r p a r t o f the boundary o f c
for a l l i,
j , and such t h a t the union o f a l l the c e l l s covers
the e n t i r e p l a n e .
A c e l l u l a r mosaic i s i l l u s t r a t e d i n F i g u r e 4 .
An a r r a y of c e l l s which is somewhat l i k e a c e l l u ­
l a r mosaic, but which v i o l a t e s the c o n v e x i t y r e ­
quirement, i s shown i n F i g u r e 5 .
Let
denote c e l l s i n a c e l l u l a r mosaic,
q is a neighbor of p if
is a curve of non­
zero l e n g t h .
It can be shown t h a t t h i s curve must
be a s t r a i g h t l i n e segment. Hence every c e l l of a
c e l l u l a r mosaic is a convex p o l y g o n .
A c e l l u l a r map is a nonempty subset of c e l l s
of a c e l l u l a r mosaic. A c e l l u l a r map may c o n s i s t
of j u s t one c e l l . Note t h a t a c e l l u l a r map need
not be connected, bounded or convex.
A chain is a sequence of c e l l s each of which
i s a neighbor o f i t s predecessor, i t s successor,
or b o t h .
A c e l l u l a r map J i s chained i f i t i s non­
empty, and i f f o r every p a i r o f c e l l s ( a , b ) i n J
t h e r e e x i s t s a c h a i n b e l o n g i n g e n t i r e l y to J and
c o n t a i n i n g c e l l s a and b.
Note t h a t if the boun­
dary of the u n i o n of the elements of a c e l l u l a r
map J is a simple closed c u r v e , then J is c h a i n e d .
A c e l l u l a r map J is the c e l l u l a r image, or
b r i e f l y the image, o f a f i g u r e p i f and o n l y i f
a) the union of the members of J c o n t a i n s p, and
b) every member of J c o n t a i n i n g an e x t e r i o r p o i n t
of p a l s o c o n t a i n s a boundary p o i n t of p. We use
the n o t a t i o n I ( p ) t o denote t h e c e l l u l a r image o f
P.
The degree of a polygon is the number of
sides it h a s .
A minimum-degree polygon of a c e l l u ­
l a r image J is any polygon p such t h a t I ( p ) = J,
and such t h a t t h e r e e x i s t s no polygon q whose
degree is l e s s than t h a t of p and such t h a t i C q ^ J .
A minimum p e r i m e t e r polygon of J is any p o l y on p such t h a t I ( p ) = J, and such t h a t t h e r e e x i s t s
no polygon q whose p e r i m e t e r is l e s s than t h a t of
p and such t h a t
The c e l l u l a r e x t e r i o r of a c e l l u l a r f i g u r e J
i s the set c o n s i s t i n g o f a l l c e l l s not i n J .
J
denotes the c e l l e x t e r i o r o f J .
boundary o f the u n i o n o f a l l c e l l s o f J .
C l e a r l y LJ is the boundary of a p o l y g o n , s i n c e
every c e l l of J is a p o l y g o n .
At each v e r t e x of
feJ draw a c i r c l e of r a d i u s e, w i t h e s u f f i c i e n t l y
s m a l l s o t h a t the c i r c l e i n t e r s e c t s o n l y the s i d e s
forming the v e r t e x .
Replace every corner of
* A c e l l u l a r mosaic i s s i m i l a r , b u t not i d e n t i c a l
t o , a " t o p o l o g i c a l complex."1
-108-
-109-
-110-
F i g . 3.
A concave blob t h a t Is both v e r t i c a l l y
convex and h o r i z o n t a l l y convex
-112-
F i g . 5.
An array of c e l l s which v i o l a t e s the
convexity requirement of a c e l l u l a r
mosaic
>114-
F i g . 6.
The minimum p e r i m e t e r p o l y g o n i n a
concave c e l l u l a r b l o b
-115-
Fig.
7.
R e l a t i o n s h i p s among v a r i o u s c l a s s e s o f
f i g u r e s and t h e i r c e l l u l a r images
-116-