In-service realtime picture quality analysis

[0002] The present invention relates to the analyzing of the quality of transmitted video images, and more particularly to an in-service, realtime picture quality analysis method that does not require a broadband reference video signal at a test point.

[0003] Currently picture quality measurements are based upon a full bandwidth picture differencing method using a human vision model algorithm, such as the Sarnoff Corporation JNDmetrix™ algorithm. In this method a reference (original) video signal is compared against a test (impaired) video signal. This requires the full bandwidth reference and test video signals to be available at a test point in order to perform picture quality analysis.

[0004] There are two ways the reference video signal may be made available at the test point:

[0005] 1. Pre-storing the reference video signal on the test system—this requires prior knowledge of reference video scenes which limits its usefulness to out-of-service and non-realtime applications; and

[0006] 2. Sending the reference video signal to the test point over a low bandwidth data link—this requires a dedicated data link to transport the reference video signal to the test point on demand as well as posing bandwidth limitation issues, and therefore is not useful for realtime picture quality analysis.

[0007] What is desired is a method of picture quality analysis that may be performed real time while the system under test is in-service.

[0008] Accordingly the present invention provides an in-service, realtime picture quality analysis method by comparing parameters that characterize the reference video signal with the same parameters for the test video signal. The reference video signal is input to a compression encoder and to a parameter characterization module. Specific parameters are extracted from the reference video signal and input to a parameter insertion module. The parameter insertion module inserts the parameters as ancillary data in the compressed data transport stream from the compression encoder. At the test point the transport stream is input to a compression decoder to recover the test video signal, and also is input to a parameter extraction module that extracts the transmitted parameters from the ancillary data. The test video signal is input to a parameter characterization module like the one at the transmission end and the same parameters are generated for the test video signal that correspond to the extracted parameters. These two sets of parameters are then input to a picture quality analyzer module for comparison, and any differences are reported as a picture quality rating.

[0009] The objects, advantages and other novel features of the present invention are apparent from the following detailed description when read in conjunction with the appended claims and attached drawing.

[0010] The FIGURE is a block diagram view of a system for performing in-service, realtime picture quality analysis according to the present invention.

[0011] Referring now to the FIGURE a broadband video signal VID_IN is input to a compression encoder 12, and is referred to as a reference video signal. The compressed video signal from the encoder 12 is transmitted via a transport stream 20 to a test point where it is input to a compression decoder 22. The output from the decoder 22 is an impaired or test video signal VID_OUT. The reference video signal also is input to a parameter characterization module 14 where certain parameters that are useful in defining the image represented by the reference video signal are generated. Such parameters may include spatial energy, temporal energy, frequency signature or the like. The parameter characterization module 14 may use one or more filters to generate the parameters, as is well known in the art. The parameters are then input to a parameter insertion module 16 which inserts the parameters as ancillary data into the transport stream 20.

[0012] At the test point the transport stream is input to a parameter extraction module 24 to extract the parameters for the reference video signal from the ancillary data. The test video signal from the decoder 22 is input to another parameter characterization module 26 that generates the same parameters for the test video signal as were generated for the reference video signal. The reference parameters from the parameter extraction module 24 and the test parameters from the parameter characterization module 26 are input to a picture quality analyzer module 28 for comparison with each other. The picture quality analyzer module 28 provides a picture quality rating PQR that represents the degradation of the test video signal with respect to the reference video signal.

[0013] For realtime analysis a relatively simple set of parameters is selected that doesn't require a lot of computation, i.e., the filters can run in real time. Thus only simple numbers are transmitted in the ancillary data, avoiding the bandwidth limitation as well as the realtime limitation. As a result the picture quality analysis algorithm implemented by the picture quality analyzer module 28 may be simpler to run in real time, as opposed to the JND algorithm used for comparing broadband video signals as in the prior art. For example the picture quality analysis algorithm may be as simple as a differencing circuit with thresholding.

[0014] Thus the present invention provides an in-service, realtime picture quality analyzer method by generating parameters from the reference video signal, transmitting them in the same transport stream with the encoded reference video signal, and then comparing them with the same parameters generated from the test video signal decoded at a test point to produce a picture quality rating.