The important visual information tends to be clustered in a small percentage of
transform coefficients, while the remaining coefficients tend to compose a dual-tree
representation. Such transform representations can be powerfully quantized and coded
with a variety of techniques depending on the available bandwidth, while a portion of
the transform-coefficient information is ignored (Zaynab et al., 2011; and Shaik and
Satyanarayana, 2012a and 2012b). The standard separable Discrete Wavelet Transform
(DWT) provides a multi-resolution depiction of a signal and has established a significant
reputation for video compression. Some recently proposed DWT-based video coders have
achieved coding competence similar to or a little better than block-based hybrid video
coders (Zhu and Ma, 2000). An important and current development in wavelet-related
research is the design and implementation of 3D multiscale transforms that represent
edges more efficiently than the Discrete Cosine Transform (DCT). This transform has
good directional selectivity (Shan and Kai-Khuang, 2000) and its sub-band responses are
around shift-invariant. Quantum Particle Swarm Optimization (QPSO) has been
implemented in a wide range of research areas such as functional optimization, pattern
recognition, neural network training and fuzzy system control, and obtained significant
success. In QPSO, a particle is an independent intelligence agent, which searches the
problem space based on its own experience and the experience of peer particles. It leads
to the problem of prematurity and easy trapping in local optimum. A modified QPSO
algorithm is proposed, that the only global best particle is agitated in every iteration of
the algorithm and other particles are updated according to the original updating
method. And this may enhance the time for reaching the best positions (Attaviriyanupap
et al., 2002). A number of variations in standard QPSO have been presented in the
literature to avoid local minimum problem. In our enhanced QPSO, at each iteration, we
are choosing ‘n’ number of best particles and distribute these best solutions with the
rest of the particles based on crossover operation. Therefore, the other particles can also
travel with the best solution instead of performing independent search. In this paper,
first the signal is decomposed by Dual-Tree real DWT (DDWT). Noise determining
scheme is used to select the significant coefficients from the DWT coefficients
(Chaturvedi et al., 2008). Later, using the enhanced version of QPSO algorithm, the Dual-
Tree sub-band coefficients, which contain high energy, are identified (Dan Simon, 2008).
The Exhaustive Search (ES) or Full Search (FS) algorithm gives the maximum Peak
Signal-to-Noise Ratio (PSNR) amongst any Block-Matching algorithm but requires
additional computational time (Ratnaweera et al., 2004; Viet-Anh and Yap-peng, 2006;
and Shaik and Satyanarayana, 2013). To reduce the computational time of ES method,
many other methods are proposed, since evolutionary computing techniques are
suitable for achieving global optimal solution (Shan and Kai-Khuang, 2000; and Viet-Anh
and Yap-peng, 2006). In this paper, we propose a QPSO algorithm to reduce the number
of computations of video compression by maintaining same or better quality of video
(Ratnaweera, 2004; Yao Nie and Kai-Khuang Ma, 2002; and Wang et al., 2010).
|