Cryptography Reference
In-Depth Information
small or zero. Only a very small portion of the errors are of large amplitude.
The error signal can be best compressed using entropy coding such as Huffman
coding or arithmetic coding. The coe
cients of a block based discrete cosine
transform (DCT) have an energy distribution that concentrates on the top
left corner of the block, and they are best represented using zigzag scanning
that starts from the top left corner.
Psychovisual Redundancy
Psychovisual redundancy refers to the information that can not be perceived
by the Human Vision System (HVS) or which the human vision system finds
insignificant. For example, the HVS has a limited response to fine spatial
detail, and is less sensitive to details near object edges or around shot-changes.
Data compression can done in such a way that information that is not per-
ceptible or of less importance is removed.
2.1.2 Evaluation of Coding Effectiveness
The effectiveness of a coding scheme can be measured by the compression
ratio. This is defined as the original uncompressed file size divided by the
resulting compressed file size. The higher the compression ratio, the smaller
the size the compressed video signal. As the compression ratio increases, the
reconstructed video signal will have bigger distortions when compared to the
original. The coding quality or fidelity decreases in this case.
There are usually three ways of measuring the coding fidelity, (1) Objec-
tive Assessment. That is, to numerically compare the pixel values before and
after coding; (2) Subjective Assessment, where human observers evaluate the
reconstructed video focusing on aesthetic acceptability; (3) Perceptual Met-
rics. These are computational algorithms that could accurately predict the
resulting scores of the subjective assessment. That is by comparison with the
human perceptual mechanism.
Objective Assessment
The Mean Square Error (MSE) and the Peak Signal to Noise Ratio (PSNR).
These are the two most commonly used objective measurements used to assess
the fidelity of the compressed video. This comparison is typically done on a
frame-by-frame basis. The MSE is the cumulative squared error between the
compressed video frame and the original frame. The PSNR is a measure of
the peak error. The MSE and PSNR are defined mathematically as,
M−1
N−1
1
MN
′
(x, y)]
2
,
MSE =
[I(x, y)−I
(2.1)
y=0
x=0
255
√
PSNR = 20log
10
MSE
,
(2.2)
