Cryptography Reference
In-Depth Information
where I(x, y) is the original video frame, I
′
(x, y) is the reconstructed frame
and M and N are the dimensions of the video frame. A small value of MSE
means lesser error and this translates to a higher value of PSNR. This is a
good quality, and a lower value represents poorer quality. Though objective
assessment is simple and straight forward, the results do not correlate well
with perceived picture quality. For example, similar PSNR in the texture
area and the smooth area of an image will exhibit different distortions when
perceived by the human eye. The HVS is less sensitive to errors in the texture
area than the smooth area.
Subjective Assessment
As the ultimate observer of the compressed video is human being, it is impor-
tant to use human observers concerning the quality of the video signal. During
subjective assessment, a number of non-expert observers are invited to such
assess the quality of the video signals. This done after the test video sequences
have been shown to them for about 10 to 30 minutes in a controlled environ-
ment. These scores cover many aspects, such as blurriness, blocking, frame
retention and jerkiness for example. According to the fidelity of the video
signal, the score can range between 1 and 5. The value 1 indicates that the
impairment is not noticeable. A value of 5 when the impairment is extremely
objectionable.
It is noted that subjective assessment reflects the perceptual nature of the
video signal under assessment. It is an extremely expensive process. It might
not be consistent as human observers are easily affected by the circumstances
which include their psychological condition, personal preferences, and viewing
condition.
Perceptual Metrics
The aim of developing a perceptual metric for video quality is to find a better
prediction for the human perception than the MSE/PSNR measure. There
have been substantial research efforts in perceptual visual quality evalua-
tion [1]. These perceptual metrics have a better performance than the tra-
ditional MSE/PNSR measure for some specific applications, it is di
cult to
find a general perceptual metric suitable for all applications. The Video Qual-
ity Experts Group (VQEG) [2] has evaluated ten perceptual metrics which
have been proposed by different researchers but none of them is statistically
better than the PSNR.
2.1.3 Classification of Video Coding Techniques
Many data compression techniques have been developed over the past three
decades. Only a few of them are suitable for application in image or video
