
192 Image and Video Compression
vide strong recovery guarantees with an efcient implementation.
In this chapter, we will consider iterative hard thresholding (IHT)
(Blumensath and Davies 2008; Portilla 2009) for discussions.
2. Iterative hard thresholding (IHT)
It is a greedy algorithm that solves the conned approximation in
the CS sparse recovery. It is formulated as
yx
x
min||||subject to |||
2
2
0
−Ψ ≤
(8.25)
This problem is solved by introducing a substitute function as follows:
xz yx xz
k
,|||||||||
|
2
2
2
2
2
2
L
S
()
=−Ψ−Ψ−Ψ+− (8.26)
The minimization of this equation under the constraint
0
ʺ
is obtained as
H=+Ψ−Ψ
(())xz
k
T
(8.