MULTIMEDIA COMMUNICATION

INSTITUTO SUPERIOR TÉCNICO

Year 2012/2013 – 2nd Semester, Responsible: Prof. Fernando Pereira

2nd Exam – 22nd June 2013 (Saturday), 9am

 

 

The marks should be out before 23rd June (Sunday), 2pm at the CMul Web page and the exam checking session will be on the 24th June (Monday), 10am in room 0.73. 

The exam is 3 hours long. Answer all the questions in a detailed way, including all the computations performed and justifying well your answers.

Don’t get ‘trapped’ by any question; move forward to another question and return later. Good luck !

 

I (1.0 + 1.0 + 1.0 = 3.0 val.)

Consider a facsimile transmission using the READ coding method at 3200 bit/s for pages with 1000 lines, each line with 1728 samples. Consider also that, on average, 75% of the samples in each line are white. Assume that

1.                the unidimensionally coded lines have an average compression factor of 15 and 25 for the black and white runs, respectively

2.                the bidimensionally coded lines have an average compression factor of 22 and 30 for the black and white runs, respectively

a)      How many bits does a unidimensionally and a bidimensionally coded line spent on average ? (R: 80.64; 62.84 bit/line)

b)     If the k parameter of the READ method is 5, what is (on average) the periodicity (in bits) of recovering the decoding synchronization? (R: 332 bits)

c)      Provide a formula for the global compression factor of a full page only as a function of the parameter k.

 

II (0.7 + 0.5 + 1.0 + 1.0 + 0.8 = 4.0 val.)

Consider the JPEG standard to code photographic images.

a)      Determine the average number of bits per pixel (considering both the luminance and the chrominances) that are spent when coding a 4:4:4 image with 8 bit/sample and a global compression factor of 20 for the luminance and 30 for the chrominances.  (R: 0.93 bit/pel)

b)     How many bits have to be spent to code a 4:4:4 colour image with 576×720 luminance resolution (8 bit/sample) if the luminance compression factor is 20 and the chrominance compression factor is twice the luminance compression factor ? (R: 331776 bit)

c)      What are the minimum and maximum numbers of non-null DCT coefficients that a JPEG block may have ? Describe the block texture content that may be associated to these minimum and maximum numbers of non-null DCT coefficients per block. (R: 1 and 64 for flat and very textured blocks, excluding pure flat black)

d)     What is the maximum number of non-null DCT coefficients that a JPEG block with pure vertical stripes may require ? What would be that same number for a JPEG block with pure horizontal stripes ? Why ? (R: 8 in the first DCT coefficients row/column)

e)      What would be effect in the decoded image if the quantization level of the DC coefficients of all blocks in the image are reduced without changing the associated quantization step ? Why ? (R: Darkening the image as the average decoding luminance would be lower)

 

III (1.0 + 1.0 + 0.5 + 1.0 = 3.5 val.)

Consider a videotelephony system using a predictive video codec for frames with 352×288 luminance resolution, 4:2:0 chrominances, 8 bit/samples, at 10 Hz. Assume that the average compression factor (measured over all macroblocks in the frame) without any restrictions on the coding modes, is 15 for the luminance and 20 for the chrominances (without considering any overhead). Also assume that the overhead amounts to 450 bits per frame.

a)      Assuming that, on average, only 200 macroblocks spend bits in each frame (the remaining macroblocks are so similar with the co-located macroblocks in the previous frame that no update is needed), determine the average compression factor (together for luminance and chrominances) measured only over the macroblocks that spend bits, considering also the overhead bits. (R: 8.21)

b)     Still in the previous situation, assume that to make the created stream more robust to errors, 1 out of each 25 (from the 200) macroblocks spending bits is mandatorily coded with the Intra mode. Assuming that the Intra mode compression factor is 60% of the compression factor found in a), determine the global compression factor (this means considering the full frame) corresponding to this Intra+Inter mix coding situation. (R: 15.83)

c)      If for editing reasons, all the macroblocks of all frames would be coded with the Intra mode, what would be the total bitrate assuming that the Intra mode compression factor would be the same as in b) ? (R: 2.47 Mbit/s)

d)     Determine the output encoder buffer size needed for the conditions in c) if a initial visualization delay lower than 200 ms should be guaranteed. Assume that the coding bits for each frame are uniformly generated in the time period between every two frames. (R: 247000 bit)

 

IV (2.5 + 0.5 = 3.0 val.)

Suppose that you are contacted by a company to design a digital storage system for short clips from the World Cup 2014. The company requires editing flexibility with a maximum access time per image below 0.8 s and needs to store the largest number of 1.5 minutes clips in a disk with 100 GBytes of capacity. The maximum access speed to the disk is 10 Mbit/s. The clips have HDTV resolution, this means 1920×1152 (Y) and 960×1152 (Cr, Cb) at 25 Hz. Assuming that you have at your disposal providing the required video quality:

1.                JPEG coding solution with a compression factor of 30 for both the luminance and chrominances

2.                MPEG-2 Video coding solution using N=6 and M=2 with the following compression factors:

a.      I frames: 20 and 25 for the luminance and chrominances, respectively

b.      P frames: 40 and 50 for the luminance and chrominances, respectively

c.      B frames: 50 and 60 for the luminance and chrominances, respectively

a)      Determine, justifying, which coding solution should be proposed to your client if the system is only for storage and non-real time playing.  (R: MPEG-2)

b)     How many full video clips would you be able to store in the disk with the two coding solutions above. (R: 300 and 410)

 

  V (0.5 + 0.5 + 1.0 + 0.5 = 2.5 val.)

Consider simple audio and video codecs such as those specified in the MPEG-1 standard.

a)      What does it happen if the ‘spatial integrity’ is lost when coding a stereo audio signal ?

b)     What does it happen if the audio encoder includes a psychoacoustic model which is conservative in the sense of lowering the hearing thresholds associated to the audio masking effects ?

c)      What does it happen in terms of compression efficiency and encoder complexity if a video encoder does not perform bidirectional motion estimation as possible in the B frames ?

d)     What does it happen in terms of random access if the video encoder only uses P and B frames ?

 

VI (1.2 + 1.0 + 1.0 + 0.8 = 4.0 val.)

Consider a DVB system for the transmission of digital TV.

a)      Regarding the audio signal, explain the benefits and drawbacks of increasing the number of audio channels, sampling frequency and number of bits/sample (1 benefit and 1 drawback for each parameter).

b)     Still regarding the audio signal, why is it advisable to adopt a frequency decomposition of the signal to efficiently code it ?

c)      Consider now that H.264/AVC is used for video coding. Why does this standard define the 4×4 and 16×16 Intra prediction modes ? For which type of content are these two prediction modes more appropriate ?

d)     Considering now the DVB-T modulation, explain which is the main benefit of increasing the number of carriers if the bandwidth and the total bitrate are kept constant. How should the number of carriers vary for larger cell sizes regarding smaller cell sizes assuming constant all the other parameters ?