Figure 4. Analog video system.

With the advent of digital television over the past 15 years, a more complex system block diagram and set of functional layers has been required as shown in Figure 5. The analog signal is converted to digital in accordance with a sampling standard such as Rec. 601. Formatting and studio interconnection of the digitized signal follow a related standard, Rec. 656, ⁷ leading to a extension in the functional layers and the variety of tests to be performed. For operational purposes, the monitoring of analog video signal properties is still key; however, this signal must be processed from the digital data. Where testing of the analog signal required only that various parameters be measured on a single waveform, digital testing requires analysis of the digital waveform, digital data formatting and digital signal coding in addition to the resulting analog signal. Again a suite of test signals is required, expanding the suite needed for analog-only testing. Although all those measurements can be performed with a signal instrument, such as the Tektronix WFM601M, there is significant processing between each pair of layers with different analysis methods for each layer as well. Prior to the advent of digital compression techniques, transmission of this higher quality signal was handled by compression back to the composite analog domain. The analog-to-digital and digital-to-analog conversion does introduce some signal quality degradation beyond that of the basic NTSC or PAL analog signal.

Figure 5. Hybrid digital/analog video system.

With the convergence of television and telecommunication, not only are there many more functional layers for the test engineers to consider but there are various possible paths with different layers. Figure 6 shows a few of the possible functional paths and layers.

Figure 6. Modern television functional layers.

Serial Digital Interconnect (SDI) is the Rec. 656 worldwide standard used for serial digital video. The SMPTE Working Group on Packetized Television Interconnections is developing a method of carrying packetized data over the same cabling and switching hardware called SDDI (for Serial Digital Data Interconnect). A networking type interconnect for the television facility, being considered by ANSI and SMPTE, is Fibre Channel which provides high speed, large packet sizes and reasonably priced hardware. The Synchronous Digital Hierarchy (SDH) telecom methods are well established worldwide and can directly carry the MPEG-2 transport stream with simple data formatting, although there is presently no standard. Looking toward the future, ATM is the expected method for transmission of packetized data, certainly for long distances and perhaps within a studio.

Three key testing layers can be defined for the modern television system as shown in Figure 7. Each has its own subset of more detailed testing layers. Video quality for compressed television systems is a much more complex matter than just using the indirect measurement methods for uncompressed video. This will be covered in detail in subsequent sections of this guide. Once the picture has been compressed, the resulting data is formatted for intra-facility connections. Examples for the use of such connections are: program interchange between video disk servers or several video/audio encoders sending single program transport streams to a multiplexer to produce a multi-program transport stream for satellite broadcasting. This is an appropriate layer for protocol testing because the data formatting can be quite complex and is relatively independent of the nature of the uncompressed signals or the eventual conversion to inter-facility transmission formats. For a majority of the television transmission systems the MPEG-2 transport stream is the common denominator at the compressed data level. The syntax and semantics for both the compressed data and the transport stream are well defined. Typical protocol testing equipment, such as the Tektronix MTS 100, will be both a source of known valid, or specifically invalid, signals and an analyzer which locates errors with respect to a defined standard and determines the value of various operational parameters for the stream of data. There are a number of possible inter-facility transmission methods as previously described. Many are well established, such as SDH/Sonet and cable television, with a variety of effective test equipment available. ATM is an emerging technology with new test equipment on the market and under development. Adaptation of traditional communication test equipment to analyze or interconnect with MPEG-2 transport streams is on the horizon.

Figure 7. Functional testing layers.

Video Quality. There are several dimensions of video quality measurement methods that need definition. These are summarized in the table below. Subjective measurements are the result of human observers providing their opinion of the video quality. Objective measurements are performed with the aid of instrumentation, manually with humans reading a calibrated scale or automatically using a mathematical algorithm.

Direct measurements are performed on the material of interest, in this case, pictures and are also called picture quality measurements. Indirect measurements are made processing specially designed test signals in the same manner as the pictures and are also called signal quality measurements. Subjective measurements are only done in a direct manner since the human opinion of test signal picture quality is not particularly meaningful. (Of course, expert viewing of full-field test signal pictures is useful as a way to determine signal distortions not for their aesthetic value.)

In-service measurements are made while the program is being displayed, directly by evaluating the program material or indirectly by including test signals with the program material. Out-of-service, appropriate test scenes are used for direct measurements and full-field test signals are used for indirect measurements.

Although there is a modest amount of compression applied to the NTSC and PAL composite systems, they are considered uncompressed in today's terminology. Signal quality (objective indirect) measurements are a reasonably good way to determine the picture quality for such uncompressed systems. That is, there is a strong mathematical correlation between subjective measurements made on pictures from the system and objective measurements made on a suite of test signals using the same system. The correlation is not perfect for all tests. There are distortions in composite systems, such as false color signals caused by high frequency luminance, which are not easily measured by objective means. Also, there are objective measurements which are so sensitive they don't directly relate to subjective results. However, such objective results are often very useful because their effect will be seen by a human observer if the pictures are processed in the same way a number of times. An example would be multiple generations using an analog video tape recorder.

The reason signal quality measurements work with analog and full-bandwidth digital systems is uncompressed systems are linear.⁸ That is, the system behavior is time invariant, signal independent and superposition applies. Signal quality measurements are made with a suite of test signals whose resulting distortions will determine transmission channel or video processing characteristics. These test signals can be very short, as an example, one line in the vertical interval. Signal quality of the uncompressed video remains critical in systems that use compression for several reasons:

• The input to a video compression codec must be accurate, in compliance with appropriate standards, and of as high a quality as possible to provide for efficient encoding.
  
  
• Video processing such as adding titles and special effects can not be accomplished in the compressed domain.
  
  
• Production facilities will not be fully compressed due to the cost and quality of compression codecs.
  
  
• The only way for different compressed formats to be interchanged is at the full bit-rate level.

This leads to a strong requirement for testing of the analog and full bandwidth digital portions as well as the sophisticated compression and transmission systems.

With the advent of compressed digital video systems the situation has become more complex. Signal quality testing will not work for the compression encoder/decoder part of the system. Traditional test signals are relatively simple compared to a natural scene and are easily compressed with little distortion or loss. Due to the ease of compression, these signals do not evaluate the encoder/decoder process. As an example, signal-to-noise ratio is not a reliable measure of picture quality, it is not a constant for a given system and it can give completely misleading results. Therefore picture quality measurements require a direct method, using natural scenes, or an equivalent thereof, which are much more complex than traditional test signals. These complex scenes stress the capabilities of the encoder resulting in non-linear distortions that are a function of the picture content.

Use of digital compression has expanded the types of distortions that can occur in the modern television system. Quantization noise which is also present in full-bandwidth digital systems is often increased by the compression system bit rate reduction process. Blockiness is a checkerboard pattern that may occur in DCT-type compression systems. Loss of resolution is common because the compression systems use the human visual system limits of acuity as a guide for removing information from the picture. Therefore, greater compression generally means less resolution. Although human acuity is less for chroma, the uncompressed picture has already used some of that latitude and compression systems often squeeze the chroma even more than the luminance. Edge busyness is another effect of quantization since more information is removed from the high-resolution parts of the picture producing noise on edges. When that noise is displaced by the compression processing into nearby flat areas it is sometimes called mosquito noise. Motion related artifacts, such as jerkiness or misplaced blocks of pixels, are present in systems which use temporal compression either based on sophisticated motion compensation or simply dropping frames because there are not enough bits available in low bandwidth systems.⁹

With the broader range of distortions to measure and the desire to optimize program distribution both technically and economically, the field of subjective measurement has expanded. Some of the subjective measurements even include an element of program quality as well as picture quality as will be discussed in detail later in this guide. Since signal quality measurements will not do the job, objective picture quality measurements are needed. Expanded types of signal quality measurements are not appropriate to cover the new subjective methods. In fact, with the increased ideas for subjective evaluation it may be true that the traditional signal quality measurements no longer have as strong a correlation with subjective requirements. There does not appear to be any plan to expand or re-test the signal quality measurement methods since there is so much work to do in developing objective picture quality methods. Such picture quality measurement methods must, also, have strong correlation with subjective measurements and cover a reasonably broad range of subjective considerations. It is expected that picture quality distortions too small for the human to see will be measured and provide an indication of the performance of concatenated systems.

Return to top