Image analysis – Pattern recognition – Feature extraction
Reexamination Certificate
2001-02-05
2004-03-16
Johnson, Timothy M. (Department: 2625)
Image analysis
Pattern recognition
Feature extraction
C382S266000, C382S274000, C382S107000, C348S014150, C348S097000, C348S100000
Reexamination Certificate
active
06707943
ABSTRACT:
FIELD OF THE INVENTION
The present invention relates to a method of monitoring the quality of distributed digital images by detecting and highlighting false contours.
BACKGROUND OF THE INVENTION
The coding methods currently employed in digital video picture transmission services have significantly reduced the quantity of information to be transmitted.
On the other hand, this reduction in the quantity of information leads to an irrecoverable loss in quality of the image as received and reproduced when compared to the source image. The magnitude of the defects generated in this way depends on the bit rate allocated to the coder and on the complexity of the image, as defined in terms of movement, brightness and texture in particular.
For technical reasons or for reasons of responsibility for the service distributed, it is necessary to evaluate continuously the quality level of the digital video signal transmitted.
Subjective evaluation methods relying on assessment by a human operative are currently widely used. They are costly and are not always totally reliable.
Consequently, it has appeared desirable to develop automatic measuring methods.
One automatic measuring method consists in differentially analysing a reference image and the image to be evaluated and allows for the human visual perception apparatus. This solution is somewhat impractical, however, because it requires the reference image to be available at the receiver.
A second feasible solution is based on an a priori knowledge of defects generated by the coding/decoding system and using statistical methods to assess the quality level of the signal by measuring the rate of occurrence of the defects.
This applies in particular to defects corresponding to a false contour effect generated over all or part of the image during the coding process. In the course of the aforementioned process, at the time of discrete cosine transform (DCT) coding, the digital image to be coded is first divided into image blocks of fixed size. The image blocks are then coded and quantised. During decoding of the image, spatial distortion between more or less perfectly reproduced blocks causes horizontal and/or vertical false contour effects, manifested completely or incompletely in the received decoded image by a spurious grid.
The techniques employed in the first solution entail lengthy and costly subjective evaluation. Also, present-day bit stream analysers, which are needed to analyse the receive images, do not provide all of the information needed to monitor the perceived visual quality and correct functioning of the service. In fact, few such analysers are capable of reflecting the perceived final quality of the received images.
Where differential analysis techniques are concerned, differential objective models require the non-coded source image to be available at the receiver, as previously mentioned. For this reason, the sequences analysed are necessarily of short duration, of the order of one second, and are therefore not representative of the service evaluated. Some defects therefore escape analysis and in particular there remain problems with synchronising the source and coded images.
Finally, because of the shortness of the sequences analysed, the analysis performed is generally discontinuous and serious errors can therefore escape analysis.
The techniques employed in the second solution appear to be even less developed.
They include the technique described in patent application EP 0 797 349 in which the object of the technique described is to highlight a block effect by gradient calculation using a Sobel filter and thereby to obtain a mask of the global and local contours. In the aforementioned method, the block effect, substantially corresponding to the false contour effect, is discriminated on the basis of false contour information which is part of the filtering window, which has a particular dimension.
SUMMARY OF THE INVENTION
One object of the present invention is to remedy the drawbacks of the prior art techniques.
Another object of the present invention is to provide a method that can be used in real time to monitor the quality of distributed digital images by detecting false contours.
Another object of the present invention is to provide a sui generis method of monitoring the quality of distributed digital images based on statistically analysing the content of usable images of a digital video sequence with no reference to any source image or to any source image coding block dimension.
Another object of the present invention is to provide a method which can be used on-line to monitor the quality of distributed digital images by detecting false contours, with no disruption of the provision of the distributed service.
Another object of the present invention is to provide a method of monitoring the quality of distributed digital images by detecting false contours that enables the provision of a distributed digital video image surveillance service.
The method in accordance with the present invention of monitoring the quality of digital images coded by blocks of pixels when the coding process generates a false contour phenomenon when the image is decoded and reproduced is noteworthy in that, for each successive current image, it calculates an image average speed vector representative of the average speed of pixels represented by at least one of their luminance, respectively chrominance, components, between said current image and the preceding image, detecting in at least one reference direction of said current image a false contour effect on the basis of a criterion discriminating the luminance, respectively chrominance, component difference between adjacent pixels of adjacent groups of pixels, and calculating a visibility coefficient of at least one current image from the value of the current image average speed vector and psycho-visual criteria relating to the existence of the false contour effect in the reference direction.
The method according to the present invention can be applied to the surveillance and management of broadcast and distributed digital video images with real time intervention.
REFERENCES:
patent: 5587927 (1996-12-01), Nagao et al.
patent: 5598482 (1997-01-01), Balasubramanian et al.
patent: 6031935 (2000-02-01), Kimmel
patent: 6185341 (2001-02-01), Ishida et al.
patent: 6438272 (2002-08-01), Huang et al.
patent: 0 797 349 (1996-09-01), None
patent: 94/09592 (1994-04-01), None
Chaddha et al., Psycho-Visual based Distortion Measures for Monochrome Image and Video Compression,Proceedings of the Asilomar Conference, N.Y. IEEE, pp. 841-845 (1993).
Lee et al., “Efficient Algorithm and Architecture for Post-Processor in HDTV”,IEEE, vol. 44, No. 1, pp. 16-26 (1998).
French Preliminary Search Report dated Oct. 2, 2000, Appl. No. FR 0002095.
Blin Jean-Louis
Gicquel Jean-Charles
Pefferkohn Stéphane
Wyckens Emmanuel
France Telecom
Johnson Timothy M.
Kassa Yosef
LandOfFree
Method of monitoring the quality of distributed digital... 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 of monitoring the quality of distributed digital..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of monitoring the quality of distributed digital... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3231362