Image analysis – Image enhancement or restoration – Edge or contour enhancement
Reexamination Certificate
1998-11-24
2001-10-23
Lee, Thomas D. (Department: 2624)
Image analysis
Image enhancement or restoration
Edge or contour enhancement
C382S236000, C348S700000, C348S701000
Reexamination Certificate
active
06307980
ABSTRACT:
The present invention deals with digital video signal processing systems, and more particularly it deals with a method of and an apparatus for introducing reference impairments into pictures to be used in subjective tests on digital video signals.
In the field of digital video signal transmission, researches are constantly made for coding (or compression) algorithms that allow reducing as much as possible the amount of information to be transmitted while guaranteeing at the same time a good quality of the coded signal. Signal quality is generally evaluated through subjective tests, in which a comparison is made between the sequence of processed pictures and a reference sequence. In case of a reasonably high quality (simply as an example, in case of coding with bit rate in the compressed sequence of the order of some Mbit/s), the reference sequence is replaced by the original unprocessed sequence. In case of compression algorithms with very low bit rates (always as an example, bit rates of the order of some kbit/s), that bring about a greater quality reduction with respect to the original sequence, it is preferred to use, as reference pictures, pictures subjected to distortion instead of those of the original sequence, so that the reference sequence quality is not exceedingly different from that of the sequence to be evaluated.
In order to guarantee a repeatability of subjective tests in time and in different measuring centres, ITU-T Draft Recommendation P.930 “Principles of a Reference Impairment System for Video” states a set of impairments to be introduced—individually or jointly—into a sequence of pictures to generate a reference sequence. The above Recommendation catalogues the main types of distortion that are observed on a video signal subjected to digital processing, and provides a brief description of the origin of such distortions.
More in detail, the distortions being taken into account in the Recommendation are as follows:
Blockiness (or block distortion): it is caused by coarse quantization of the spatial frequency components during the coding process and it is generally more visible in the smoothest picture areas located next to moving edges;
Blurring: it is the reduction in sharpness of edges and in spatial detail caused by the need of a trade off, in compression algorithms, between number of usable bits on the one hand and resolution provided by code and motion representation on the other hand;
Edge Busyness (echoes or pixel replicas with time-varying positions and intensities): it is a distortion that is concentrated at object edges, and it is caused by the use of relatively coarse quantization levels when coding a block that includes both edges delimiting rather smooth areas and pixels with a quite different average luminance level;
Noise around moving edges (known as Mosquito noise): it is a noise substantially similar to the previous one, characterised by added moving artefacts or by blotchy patterns superimposed over the edges;
“Pepper and salt” effect or quantization noise: it is a typical noise of the digital conversion processes and its appearance is similar to the well-known “snow effect” of the analogue television, though it is distributed over the picture in a non-uniform way;
Jerkiness: as its name states, it consists of the perception of an originally smooth movement as a series of snapshots.
An Appendix to the Recommendation also discloses the way of simulating the above distortions as regards luminance, and includes a proposal to implement a system to generate them. This known system operates as follows:
blockiness: it is obtained by identifying picture edges, to recognise areas where such distortion is more commonly visible; in these areas, a certain number of blocks of N×N pixels where the distortion must be introduced are selected, through an analysis of the displacement characteristics with respect to the previous frame; to actually introduce the distortion, the pixel luminance value in each selected block is then replaced by a new value obtained by adding a random value ranging between −2 and +2 to the average between the original pixel luminance and the average block luminance; every block keeps this distortion for a fixed number of frames (for example 15) to make this effect more easily perceivable;
blurring: it is implemented by applying, frame by frame, a low-pass unidimensional digital filtering on each frame line and by multiplying the filtered samples by a normalisation coefficient, in order to recover the correct amplitude dynamics;
edge busyness: it is implemented through a bidimensional filtering with filters having ripples in the passband amplitude response: the ripple amplitude determines the amplitude or intensity of the echoes that are added to simulate distortion, the ripple frequency determines the echo displacement with respect to the picture and the phase determines the echo polarity;
quantization noise: it is simulated by randomly choosing a pre-set number of pixels and by replacing the luminance value with a random value within a range of allowed values;
mosquito noise: it is simulated by adding a random value, chosen within pre-set a range, to the luminance value for each pixel classified as belonging to edges and having an associated motion;
jerkiness: it is simulated by repeating a same frame for a certain number of times.
The tests carried out have proven that the system proposed in the Appendix to the Recommendation is not fully satisfactory—that is, it does not allow an easy distortion recognition—at least as regards edge busyness and blockiness distortions.
According to the present invention, instead, a method and an apparatus are provided through which the above distortions are generated so that the perception thereof is made easier during the tests.
The features of the invention are disclosed in the following claims
1
to
12
, as regards the method, and in claims
13
to
15
, as regards the apparatus.
REFERENCES:
patent: 5367343 (1994-11-01), Blair
patent: 5701163 (1997-12-01), Richards et al.
patent: 6005638 (1999-12-01), Blair et al.
patent: 6181828 (2001-01-01), Hogan
XP-000721388 —A Generalized Block-Edge Impairment Metric for Video Coding, pp. 317-320, vol. 4, No. 11, Nov. 1997.
Cselt-Centro Studi e Laboratori Telecomunicazioni S.p.A.
Dubno Herbert
Lee Thomas D.
LandOfFree
Method and apparatus for introducing reference impairments... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method and apparatus for introducing reference impairments..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method and apparatus for introducing reference impairments... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2569173