Convex Hulls

algorithms and data structures

• Convexivity
• Package-Wrap Algorithm
• Graham Scan
• Dynamic Convex Hull

Convex Hull 1
 What is a Convex Hull?
• Let S be a set of points in the
plane.
• Intuition: Imagine the points of
S as being pegs; the convex
hull of S is the shape of a
rubber-band stretched around
the pegs.
• Formal definition: the convex
hull of S is the smallest convex
polygon that contains all the
points of S.

Convex Hull 2
 Convexity
convex

• A polygon P is said to be
convex if:
– P is non-intersecting; and
– for any two points p and q Eh? WhatÕs
on the boundary of P, convex?
segment pq lies entirely
inside P

Non
convex
Convex Hull 3
 Why Convex Hulls?
Who cares about
convex hulls?
I donÕ t ...
... but robots do!

shortest path
avoiding the
obstacle

obstacle

Convex Hull 4
 The Package Wrapping
Algorithm
Idea: Think of wrapping
a package. Put the paper
in contact with the
package and continue to
wrap around from one
surface to the next until
you get all the way
around.

Convex Hull 5
 Package Wrap
• Question: Given the current point, how do we compute the
next point to be wrapped?
– Set up an orientation tournament using the current point as the
anchor-point.
– The next point is selected as the point that beats all other points at
Counter Clockwise(CCW) orientation, i.e., for any other point, we
have orientation(c, p, q) = CCW
q

Convex Hull 6
p
c
 Time Complexity of Package Wrap
• For every point on the hull we examine all the other points to
determine the next point
• Notation:
N: number of points
M: number of hull points (M  N)
• Time complexity: (M N)
• Worst case: (N2)
(All the points are on the hull, and thus M = N)
• Average case: (N log N ) — (N4/3)
for points randomly distributed inside a square, M = (log N) on average
for points randomly distributed inside a circle, M = (N1/3) on average

Convex Hull 7
 Package Wrap has worst-case
time complexity O( N2 )
• Which is bad...

N2
But in 1972, Nabisco needed a better
cookie - so they hired R. L. Graham, who
came up with...
Convex Hull 8
 The Graham Scan Algorithm
• Rave Reviews:
– “Almost linear!”
Sedgewick ÒAbetter crunch!Ó
– “It’s just a sort!” Atul
– “Two thumbs up!” Siskel
and Ebert

Nabisco says...

and history was made.

Convex Hull 9
 Graham Scan
• Form a simple polygon (connect the
dots as before)
• Remove points at concave angles
(one at a time, backtracking one step
when any point is removed).
• Continue until you get all the way
around.

Convex Hull 10
Graham Scan How Does it Work?

• Start with the lowest point (anchor point)

Convex Hull 11
 Graham Scan: Phase 1
• Now, form a closed simple path traversing the points by
increasing angle with respect to the anchor point

Convex Hull 12
 Graham Scan: Phase 2

• The anchor point and the next point on the

Convex Hull 13
 Graham Scan: Phase 2
• keep the path and the hull points in two sequences
– elements are removed from the beginning of the path sequence and
are inserted and deleted from the end of the hull sequence
• orientation is used to decide whether to accept or reject the
next point

next
cur
Convex Hull
prev 14
 n (p,c,n) is a

c right turn!
p

Discard c

c
p
Convex Hull 15
 n p
(p,c,n) is a
c right turn!

p
n c
(p,c,n) is a
right turn!

n c
p

Convex Hull 16
Time Complexity of Graham Scan
• Phase 1 takes time O(N logN)
points are sorted by angle around the anchor
• Phase 2 takes time O(N)
each point is inserted into the sequencen exactly once,
and each point is removed from the sequence at most
once
• Total time complexity O(N log N)

Convex Hull 17
 How to Increase Speed
• Wipe out a lot of the points you know won’t be on the hull! This is
called interior elimination. One approach:
– Find the farthest points in the SW, NW, NE, and SE directions
– Eliminate the points inside the quadrilateral (SW, NW, NE, SE)
– only O(√N) points are left on average, if points are uniformly distributed on a
unit square!
• Do Package Wrap or Graham Scan on the remaining points
NW
NE

SW SE
Convex Hull 18
 Dynamic Convex Hull

• The basic convex hull algorithms were fairly interesting, but you
may have noticed that you can’t draw the hull until after all of the
points have been specified.
• Is there an interactive way to add points to the hull and redraw it
while maintaining an optimal time complexity?

Convex Hull 19
 Two Halves = One Hull
• For this algorithm, we consider a convex hull as being two
parts:

An upper hull...
B

and a lower hull...

Convex Hull 20
 Adding points: Case 1
• Case 1: the new point is within the horizontal span of the hull
Upper Hull 1a:
If the new point is above the upper hull, then it should be added to the upper hull
and some upper-hull points may be removed.

.
Upper Hull 1b:
If the new point is below the upper hull, no changes need to be made

Convex Hull 21
 Case 1 (cont.)
• The cases for the lower hull are similar.
Lower Hull 1a:
If the new point is below the lower hull, then it is added to the lower
hull and some lower-hull points may be removed.

Lower Hull 1b:

Convex Hull 22
 Adding Points: Case 2

• Case 2: the new point is outside the horizontal span of the

Convex Hull 23
 Hull Modification
• In Case 1, we determine the vertices l and r of the upper or lower
hulls immediately preceding/following the new point p in the x-
order. l r

If p has been added to the upper hull, examine the upper hull
rightward starting at r. If p makes a CCW turn with r and its right
neighbor, remove r. Continue until there are no more CCW turns.
Repeat for point l examining the upper hull leftward. The computation
for the bottom hull is similar CCW

Convex Hull
Bad 24