TWO MARK QUESTIONS AND ANSWERS 1. What is rate distortion distortion theory? Rate distortion theory is concerned with the trade-offs between distortion and rate in lossy compression schemes. If the average number of bits used to represent each sample value i.e. the rate is decreased there will be an increase in distortion. This is rate distortion theory. 2. Define the asi! !on!e"ts of infor#ation infor#ation theory. If the probability of an event is low the amount of self information associated with it is high; i.e. the probability of an event is high, the information associated with it is low. The self information information associated associated with any any event is is given by, i (A! -log p(A. $. Write any three te!hni%&es for 'oss'ess !o#"ression? i "uffman coding. ii #hannon fano coding. iii Arithmetic coding. (. )o* entro"y is re'ated for the "erfor#an!e #eas&res? The entropy entropy is a measure measure of the average average number number of binary binary symbols needed to code the o$p of the source. "ence for a compression scheme to be lossless it is necessary to code the o$p of the source with an average number of bits e%ual to the entropy of the source. +. What do yo& #ean y 'ossy and 'oss'ess !o#"ression? If the reconstructed data in the received end is same as that of the original data, then it is a lossless compression. If the reconstructed data in the received end is differs from that of the original data, then it is a lossy compression. ,. Write any three te!hni%&es for 'ossy !o#"ression? i #ubband coding ii &avelet based compression iii ')* -. Define e!tor %&anti/ation and 0ie its #erit oer s!a'ar %&anti/ation. If the set of inputs + outputs of a %uantier are vectors then it is called vector
%uantiation. or a given rate the use of vector %uantiation results in a lower distortion than scalar %uantiation. . What are the i#"ortant a""'i!ations of data !o#"ression? ata compression schemes find its application in mobile communication, digital T/, satellite T/. T/. . Write the ta3ono#y of !o#"ression te!hni%&es? 0ased on the re%uirements of reconstruction, data compression scheme can be classified as lossy compression and lossless compression. compression. 14. What is #eant y !o#"anded %&anti/ation? In companded %uantiation the input is first mapped through a compressor function. This function stretches the high probability regions close to the origin and correspondingly compresses the low probability regions away from the origin. The output of the compressor function is %uantied using a uniform %uantier + the %uantied value is transformed via an e1pander function. This is 2nown as companded %uantiation. 11. What is #eant y #ode''in0? 3odelling is nothing but e1tracting the information about any redundancy that e1ist in the data and describing the redundancy in the form of a model. 12. What are the "arts of h&#an a&dio is&a' syste#? The various various parts of a human human audio audio visual system system are retina, retina, fovea, tympanic, membrane, cochlea, oval window etc. 1$. 5ie so#e #ode's that are &sed in 'oss'ess !o#"ression? The various various models that that are used in a lossless compression compression scheme scheme are probabilistic models, physical models, and mar2ov models, composite source models etc. 1(. 5ie so#e #ode's that are &sed in a 'ossy !o#"ression. The various various models that that are used in a lossy compression compression scheme scheme are probabilistic models, physical models, and linear system models. 1+. What is a !o#"osite !o#"osite so&r!e #ode'? In many applications, it is not easy to use a single model to describe the source. In such cases, we can define a composite source , with only one source being active at a time. 1,. What are "refi3 !odes?
A code in which no code word is a prefi1 to another code word is called prefi1 code (eg."uffman code 1-. 5ie any t*o !hara!teristi!s !hara!teristi!s of a !ode. i A code should be uni%uely decodable. ii The code words for letters that occur more fre%uently are shorter than for letters that occurs less fre%uently. 1. What are the t*o ty"es of %&anti/ation error? *ranular error and slope over load error. 1. What are t*o ty"es of ada"tie %&anti/ation? %&anti/ation? The two types types of adaptive %uantiati %uantiation on are forward forward adaptive adaptive %uantiation and bac2ward adaptive %uantiation. 24. What do yo& #ean y for*ard ada"tie %&anti/ation? In forward adaptive %uantiation, the source output is divided into bloc2s of data. )ach bloc2 is analyed before %uantiation, and the %uantier parameters are set accordingly. The settings of the %uantier are then transmitted to the receiver as side information. 21.What is #eant y o"ti#&# "refi3 !odes? In an optimum code, symbols that occur more fre%uently (have a higher probability of occurrence will have shorter code words than symbols that occur less fre%uently. In an optimum code, the two symbols that occur less fre%uently will have the same length. 45. Write any three te!hni%&es for 'oss 'ess !o#"ression? a. "uffman coding b. Adaptive "uffman coding c. Arithmetic coding d. #hannonfano coding 22. What are the a""'i!ations of Arith#eti! !odin0? a. 0i-level Image compression-'0I* standard b. '0I*4 c. Image compression di0ra#6!odin0 #ean? 46. What does di0ra#6!odin0 a. It is one of the most common forms of static dictionary coding. b. In this form of coding, the dictionary consists of all letters of the source alphabet followed by as many pairs of letters called digrams. •
•
c. The digram encoder reads two character input and search the dictionary to see if this input e1ists. If so, inde1 is encoded and transmitted. 2(. What is 78 fa#i'y a'0orith#? a. It is one of the widely used adaptive dictionary based techni%ue. b. 7899, 789:, 78& are the different approaches in this algorithm. 2+. Define offset in 78-- a""roa!h? To encode encode the se%uence se%uence in the loo2 loo2 ahead buffer, buffer, the encoder encoder moves a search pointer bac2 through the search buffer until it encounters a match to the first symbol I the loo2 ahead buffer. The distance of the pointer from the loo2 ahead buffer is called the offset. 4. Define sear!h &ffer and 'oo9 ahead &ffer? The encoder encoder e1amines e1amines the input se%uence se%uence through through a sliding sliding window. window. The window consists of two parts, search buffer that contains a portion of the recently encoded se%uence and a loo2 ahead buffer that contains the ne1t portion of the se%uence to be encoded. 2-. 5ie so#e a""'i!ation of 78W? a. ile compression-<=I> compression b. Image compression-*I (*raphics Interchange ormat c. ?ompression over 3@)3#-/.4 bits 4:. What does Stati! Di!tionary #ean? a. #tatic dictionary are used when prior 2nowledge about the sources available. b. epending on the input source the dictionary is adapted. 2. What is #eant y ada"tie di!tionary? a. This is used in situation when there is no prior 2nowledge about the sources available. b. epending on the input source the dictionary is adapted. 6B. Is )&ff#an !odin0 'ossy or 'ess !o#"ression? "uffman coding is a loss less compression. $1. 5ie an e3a#"'e for off'ine en!odin0 and on'ine en!odin0? a. "uffman coding is a offline coding in which the dataCs are stored in a buffer and then encoded. b. Adaptive "uffman coding is an online encoding in which the input data coded directly by using tree encoding techni%ue. 64. Define 'e3i!o0ra"hi! orderin0? In le1icographic ordering the ordering of the letters in an alphabet induces an ordering on the words constructed from this alphabet. )1D ordering of words in a dictionary.
$$. What is the a'0orith# &sed in :;I5? a. rogressive transmission algorithm b. 7oss less compression algorithm. $(. What are the different a""roa!hes a""roa!hes &sed in a ada"tie di!tionary te!hni%&e? a. 7899 b. 789: c. 78& $+. What is #eant y "ro0ressie trans#ission? In progressive transmission of an image, a low-resolution representation of the image is sent first. This low resolution re%uires only few bits to encode. The image is then updated or refine to the desired fidelity by transmitting more and more information. $. Define 'oss'ess !hanne'. The channel channel described described by a channel channel matri1 matri1 with only one one nonero nonero element in each column is called a lossless channel. In the lossless channel no sources information is lost in transmission. $. Define Deter#inisti! Deter#inisti! !hanne' A channel described by a channel matri1 with only one nonero element in each row is called a deterministic channel and this element must be unity. (4.What is the i#"ortant of s& and !odin0 in a&dio !o#"ression? In sub band coding the source output is separated into different bands of fre%uencies. This results in fre%uency bands with different characteristics. Thus we can choose the compression scheme most suited to that characteristic. characteristic. It also helps to a variable bit allocation to the various fre%uency components depending upon the information content. This decreases the average number of bits re%uired to code the source output. (1. What are the "ara#eters &sed in si'en!e !o#"ression? #ilence compression in sound files is the e%uivalent of run length coding on normal data files. The various parameters are A threshold value below which it can be considered as silence •
A special Esilence codeC followed by a single byte that indicates how many consecutive silence codes are present A threshold to recognie the start of a run of silence. @nly if we have sufficient bytes of silence we apply silence coding. A parameter to indicate how many consecutive non-silence codes are needed, after a string of silence, so that we can declare the silence runs to be over. (2. 7ist the ario&s ana'ysis < synthesis s"ee!h s!he#es. The various various analysis $ synthesis synthesis schemes schemes are ?hannel vocoders (each segment of input speech is analyed using a ban2 of filters 7inear predictive vocoders (<# govt. standard at the rate of 4.Fbps ?ode e1cited linear prediction (?)7 based schemes. Gfederal standard 5B5 (#-5B, *.94: speech standardH #inusoidal coders which provide e1cellent performance at rates of .:Fbps and 3i1ed e1citation linear prediction (3)7, Gwhich is the new standard 4.Fbps federal standard speech coderH. ($. What are the fa!tors to e !onsidered for a oi!ed< &noi!ed de!ision in "redi!tie !oders? The following following factors are are considered considered /oiced speech (a$e$o has larger amplitude and hence more energy than unvoiced signals(s$f
•
•
• •
•
•
•
•
•
7ayer 5 + 4 coders both use a ban2 of 64 filters, splitting the input into 64 bands. )ach band has a bandwidth of fs$, where fs is the sampling fre%uency. Allowable sampling fre%uencies are 64, BBB, , 5BB, :,BBB samples$sec. The output output of the sub band band is %uantied %uantied using a uniform %uantiser. %uantiser. A variable length %uantiser is used. The number bits are assigned based on the mas2ing property of human ears. That is, if we have a large amplitude signal at one fre%uency, the audibility of the neighboring signals is affected. "ence if we have a large signal in one of the sub bands, we need only fewer bits to code a neighboring sub band. (+. Define o!oders and *hat are the ty"es !hanne' o!oders? /ocoders are also called voice coders. /ocoders reproduce synthetic sounding which is somewhat artificial %uality. They can transmit signals at a lower bit rate in the range of 5.4 to 4.Fb. The receiver uses the model parameters along with the transmitted parameters to synthesie the appro1imation to the source output. The types of of channel vocoders vocoders are linear linear predictive predictive coder coder and code code e1cited linear prediction. (,. What is 9no*n as %&adratie #irror fi'ter? The filter ban2 ban2 in sub band band coding coding consists of a cascade cascade of filter filter stages where each stage consists of 7 + ". The most popular among the filter is 3. These filters have the property that the impulse response of a 7 is given by hn. Then the high pass impulse response is given by G(-5Jnh=-5-nH. Gh=-5-nH ! hn, n!B,5,-------(=$4-5. n!B,5,-------(=$4-5. (-. A""'i!ation to s"ee!h !odin0 5.-22 *.944 provides a techni%ue for wide band coding of speech signals that is based on sub band coding. They provide high %uality speech at Fbps. The two other models used to encode the input are at K + :Fbps. These two models are used when au1iliary channels are used. (. Define it a''o!ation?
The allocation allocation of bits between between the sub sub bands is an important important design parameter. ifferent sub band contain different amounts of information. &e need to allocate the available bits among the sub bands according to some measure of information. The bit allocation allocation can have have a significant significant impact impact in the %uality %uality of the final reconstruction especially when the information content of different bands is very different. (. 5ie an e3a#"'e of it a''o!ation "ro!ed&re in asi! s& and !odin0? #uppose we are decomposing the source output into four bands and we want a coding rate of 5 bit per sample. &e can accomplish this by using 5 bit per sample for each of the four bands. @therwise we could simply discard the output of two of the bands and use 4 bits per sample for the two remaining bands. @ther way is to discard the output of three of the four filters and use bits per sample to encode the output of the remaining filter. +4. What is #eant y de!i#ation > inter"o'ation? #uppose we have a se%uence >B, >5, >4,----, then we can divide the se%uence into two sub se%uences, i.e., >B, >4, >, ------+ >5, >6, -----w here 5$ corresponds to a delay of one sample and 3 denotes sub sampling by a factor of 3. this sub sampling process is called down sampling or decimation. The original original se%uence can be recovered from the two down sampling by inserting eros between consecutive samples of the sub se%uences. This is called up sampling or information. +1. The !&toff fre%&en!y for anti6a'iasin0 fi'ter is - K)/ and K)/. 5ie reasons. )ven though the speech output is sampled at 5,BBB samples per second, the cutoff fre%uency for the anti-aliasing filter is 9 F". The reason is that the cutoff fre%uency for the anti-aliasing filter is not going to be sharp li2e that of the ideal
low pass filter. Therefore, the highest fre%uency component in the filter output will be greater than 9 F". +2. )o* #as9in0 "ro"erties of the h&#an ear &sed I red&!in0 the n&#er of its in &nifor# %&antiser? If we have a large amplitude signal at one fre%uency, it affects the audibility of signals at other fre%uencies. In particular, a loud signal at one fre%uency may ma2e %uantiation noise at other fre%uencies inaudible. Therefore, Therefore, if we have a large signal in one of the sub bands, we can tolerate more %uantiation error in the neighboring bands and use fewer bits. #o number of bits can be reduced in uniform %uantiser. +$. 5enera''y a&to !orre'ation f&n!tion is &sed as a too' for otainin0 the "it!h "eriod. ;&t in 'inear "redi!tie !oders AMD@ is &sed. Why? /oiced speech is not e1actly periodic which ma2es the ma1imum lower than we would e1pect from the periodic signal. *enerally, a ma1imum is detected by chec2ing the auto correlation value against the threshold. If the value is greater than the threshold, a ma1imum is declared to have occurred. &hen there is uncertainty about the magnitude of the ma1imum value, it is difficult to select a value for the threshold. Another problem occurs because of the interference due to other resonances in vocal tract. #o Average 3agnitude ifference unction (A3 is used. +(. What are for#ants? What are its "ro"erties? All fre%uency components of speech are e%ually important. As the vocal tract is a tube of non uniform cross section, it resonates at a number of different fre%uencies. These fre%uencies are called formants. The formants formants values change change with different different sounds, sounds, but ranges ranges in which it occurred can be identified. or e1ample, the first formant occurs in the range 4BBB-:BB " for a male spea2er and in the range 4KB-5BBB " for a female spea2er. ++. What is re0&'ar "&'se e3!itation !odin0?
The 3ulti ulseulse- 7inear 7inear rediction rediction ?oding (3-7? (3-7? algorithm algorithm was was later modified. Instead of using e1citation vectors in which the nonero values are separated by an arbitrary number of ero values, they forced the nonero values to occur at regularly spaced intervals. urthermore, they allowed the nonero values to ta2en on a number of different values. This scheme is called regular pulse e1citation (R) coding. +,. Define a'iasin0. What is anti6a'iasin0 fi'ter? ?omponents with fre%uencies higher than half the sampling rate show up at lower fre%uencies. This is called aliasing. In order to prevent aliasing, most systems that re%uire sampling will contain an anti-aliasing filter that restricts the input to the sampler to be less than half the sampling fre%uency. +-. 5ie the Ny%&ist r&'e. If the highest fre%uency component of a signal is fB, then we need to sample the at a fre%uency more than 4fB times per second. This result is 2nown as the =y%uist theorem or =y%uist rule. It can also be e1tended to signals that only have fre%uency components between two fre%uencies f5 and f4. In order to recover the signal e1actly, we need to sample the signal at a rate of at least 4(f4-f5 samples per second. +. What are fi'ter !oeffi!ients? !oeffi!ients? The general general form of the input-output input-output relationships relationships of the filter filter is given by =3 Ln ! M ai >n-I N M bi Ln-i i!B i!5 where the se%uence O>nP is the input to the filter, the se%uence OLn P is the output from the filter, and the values OaiP and ObiP are called the filter coefficients. +. What is @IR and IIR fi'ters? If the input se%uence is a signal 5 followed by all Bs, the output se%uence is called the impulse response of the filter. If the filter coefficient bi are all B, then the impulse response will die out after = samples. These filters are called inite Impulse Response (IR filters. The
number = is sometimes called the number of taps in the filter. If any of the bi have non ero values, the impulse response can continue forever. ilters with non ero value for some of the bi are called Infinite Impulse Response (IIR filters. ,4. Define %&ad tree? uad tree is a portioning method in fractal compression. In this method we start by dividing up the image into ma1imum sie range bloc2s. If a particular bloc2 does not have a satisfactory reconstruction we can divide it up in to bloc2s. These bloc2s in turn can also if needed be divided in to bloc2s. This method of partitioning is called %uad tree partitioning ,1. In *hat *ay S=I)T is #ore effi!ient than E8W? In )8& when entire sub tree is in significant we transmit all coefficients in it with a ero tree root label r The #I"T algorithm algorithm uses uses portioning portioning of trees in such a manner manner that it tend to 2eep insignificant coefficients together in larger subsets. ,2. What is #assi! transfor#ation? transfor#ation? In fractal compression massic transformation adQusts the intensity and orientation of pi1els in domain bloc2. trans#ission? 6. What is "ro0ressie trans#ission? In progressive transmission of an image a low resolution representation of image is sent first. This low-resolution representation re%uires very few bits to encode. ,(. What is the differen!e in :=E5 and :=E5 2444? ')* 4BBB differs differs mainly by means of transform transform coding. coding. In ')* 4BBB wavelets are used to perform decomposition of image. In ')* ?T is used ,+. What is fra!ta' !o#"ression? !o#"ression? fractal compression is a lossy compression method used to compress images using fractals. The method is best suited for photographs of natural scenes. fractal compression techni%ue relies on the fact that in certain images parts of images resembles other part of same image. ,,. What is E;OT? )mbedded 0loc2 ?oding with @ptimied Truncation is a bloc2 coding scheme. It generates an embedded bit stream. It organies bit steam in a succession of layers. )ach layer corresponds to certain distortion level. &ith in
each layer each bloc2 is coded with a variable number of bits. The partitioning of bits between bloc2s is obtained using a lagrangian optimiation that dictates the partitioning. The %uality of reproduction is proportional to noD of layers received. 9. What is a *ae'et transfor#? &avelets are functions defined over a finite interval and having an average value of ero. 0asic idea of wavelet transform is to represent any arbitrary function as a super position of a set of such wavelets or basic functions. These basic functions are obtained from a single prototype wavelet called mother wavelet by dilations or contractions and translations. ,. Define de'ta #od&'ation elta modulation is the one-bit version of differential pulse code modulation. ,. Define ada"tie de'ta #od&'ation The performance performance of a delta delta modulator modulator can be be improved improved significantly significantly by ma2ing the step sie of the modulator assume a time- varying form. In particular, during a steep segment of the input signal the step sie is increased. ?onversely, when the input signal is varying slowly, the step is reduced , In this way, the step sie is adapting to the level of the signal. The resulting method is called adaptive delta modulation (A3. -4. Na#e the ty"es of &nifor# %&anti/er? 3id tread type %uantier. 3id riser type %uantier. -1. Define #id tread %&anti/er? @rigin of the signal lies in the middle of a tread of the staircase. @utput 6a 4a a @verload level -6a$4 -a$4 Input -a -4a ea2 to pea2 e1cursion where a!delta -2. Define %&anti/ation error? uantiation error is the difference between the output and input values of %uantier.
-$. Define #id6riser #id6riser %&anti/er? @rigin of the signal lies in the middle of a riser of the staircase @$p 6a$4 a$4 @ver load level a 4a 6a a i$p -(. Dra* the %&anti/ation error for the #id tread and #id6rise ty"e of %&anti/er? For mid tread typeD uantiation error a$4 Input -a$4 a For mid riser type: uantiation error a$4 Input A -+.What yo& #ean y non6&nifor# %&anti/ation? #tep sie is not uniform. =on uniform %uantier is characteried by a step sie that increases as the separation from the origin of the transfer characteristics characteristics is increased. =on-uniform %uantiation is otherwise called as robust quantization -,. What is the disadanta0e disadanta0e of &nifor# %&anti/ation oer the non6&nifor# %&anti/ation? #=R decreases with decrease in input power level at the uniform %uantier but non-uniform %uantiation maintain a constant #=R for wide range of input power levels. This type of %uantiation is called as robust %uantiation. --.What is ideo !o#"ression? /ideo compression can be viewed as the compression of a se%uence of images; in other words image compression with temporal component. "ere the video compression ma2es use of temporal correlation to remove redundancy -.What is #otion !o#"ensation? The previous previous reconstructed reconstructed frame is is used to generate generate a prediction for the current frame, the prediction error or residual, is encoded and transmitted to the receiver. The previous reconstructed frame is also available at
the receiver. #o the receiver 2nows the manner in which the prediction was performed, it can use this information to generate the prediction values and add them to the prediction error to generate the reconstruction. This prediction operation in video coding has ta2en to account the motion of obQects in frame, which is 2nown as motion compensation -.What are the disadanta0es disadanta0es of ideo !o#"ression? &e do not perceive the motion video in the dame manner as we perceive the still images. 3otion video may mas2 coding aircrafts that would be visible in still images. @n the other hand aircrafts that may not be visible in reconstructed still images can be very annoying in the reconstructed motion video se%uences 4.What is the adanta0e of 'oo" fi'ter? #harp edges in the bloc2 used for prediction can result in generation of sharp changes in the prediction error. This in turn causes high values for high fre%uency coefficients in transforms, which can increase transmission rate. To avoid this prior to ta2ing the difference, the prediction bloc2 can be smoothed by using a two dimensional spatial filter. The filter is separable, it can be implemented as a one dimensional filter that first operates on rows and then on columns. The filter coefficients are , S, , e1cept at bloc2 boundaries where one of the filter taps would fall outside bloc2 1.Differentiate 1.Differentiate 0'oa' #otion > 'o!a' #otion? In three dimensional model based approach to compression of facial image se%uences, a generic wire frame model is constructed using triangles. @nce this model is available to both transmitter and receiver, only changes in faces are transmitted to receiver. These changes can be classified as global motion or local motion. global motion involves movement of head while local motion involves changes in features i.e. changes in facial e1pression. 2.What is M=E56(?
The standard standard views a multimedia multimedia scene scene as a collection collection of obQects. obQects. These obQects can be visual such as still bac2ground or a tal2ing head such as music, speech and so on. )ach of these obQects can be coded independently using different techni%ues to generate separate elementary bit streams. These bit streams are multiple1ed along with a scene description. the protocol for managing the elementary streams and their multiple1ed version called )7I/)RL 3<7TI3)IA I=T)*RATI@= RA3) &@RF is an important part of 3)*- $. What is ).2,1 standard? The earliest earliest ?T based video video coding coding standard standard is ".45 standard. standard. An i$p image is divided in to bloc2s of :: pi1els. or a given :: pi1el we subtract the prediction generated using the previous frame. The difference btw the bloc2 being encoded and the prediction is transformed using a ?T. The transform coefficients are %uantied and the %uantiation label encoded using a variable length code. (. What is M=E5? #tandards for asymmetric applications have been developed by I#@ + I)?, which is 2nown as 3)*. 3)* was set up at different rates for applications that re%uire storage of audio and video on digital storage media. 3)*5 3)*4+ 3)*6 are targeted at rates of 5.K, 5B,B3b$s. respectively. +. 7ist o&t one a""'i!ation of *ae'et ased !o#"&tation? @ne of the most popular applications of wavelet has been to image compression. The ')*4BBB standard is designed to update and replace the current ')* standard will use wavelets instead of ?T to perform decomposition of image. ,. What is a #other *ae'et? The functions functions that are obtained obtained by by changing changing the sie if function function or scaling and translating the single function are called mother wavelet. -. What is 0ro&" of "i!t&res? ifferent frames li2e 0 frames, I frames, frames are organied together in a group . This group is called as group of pictures. *roup of pictures is a small random access unit in the video se%ence. . 5ie the different orders in M=E51 standarad.
isplay order and bitstream order isplay order is the se%uence in which video se%uence is displayed to the user 0itstream order is the processing order that is different from the display order. . What is !onstrained "ara#eter "ara#eter itstrea#? The 3)* committee committee has has provided some some suggested suggested values for the the various parameters li2e vertical picture sie ,horiontal picture sie ,pi1el rate.These suggested values are called constrained parameter bitstream. 4. What do yo& #eant y "rofi'es and 'ee's in M=E5 2 standard? rofile defines the algorithm that is used in 3)* 4 standard and levels defines the constraints on the parameters. 1. Na#e the "rofi'es and 'ee's &sed in M=E5 2 standard. rofiles are simple, main, snr-scalable, spatially scalable, and main. The levels are are low, main, high 5B and high. The low level level corresponds corresponds to a frame sie sie of 6K4 1 4B 4B The main level level corresponds corresponds to a frame frame sie of 94B 94B 1 :B The high 5B 5B level correspon corresponds ds to a frame sie sie of 5B 1 55K4 55K4 The high level level corresponds corresponds to a frame frame sie of 5U4B 5U4B 1 5B:B 2. What is DMI@? 3I means elivery 3ultimedia Integration ramewor2. It is a protocol for managing the elementary bit streams and multiple1ed version of different bit streams. It is used in 3)* standard. $. What is a "ost dete!tion fi'ter? The post detection detection filter named asV base-band base-band low low pass filterV has has a bandwidth that is Qust large enough to accommodate the highest fre%uency component of the message signal. (. What is the disadanta0e disadanta0e of &nifor# %&anti/ation oer the non6&nifor# %&anti/ation? #=R decreases with decrease in input power level at the uniform %uantier but non-uniform %uantiation maintain a constant #=R for wide range of input power levels. This type of %uantiation is called as robust %uantiation. +. What are the t*o fo'd effe!ts of %&anti/in0 "ro!ess.
The pea2-to-pea2 pea2-to-pea2 range range of input input sample values values subdivided subdivided into a finite finite set of decision levels or decision thresholds .The output is assigned a discrete value selected from a finite set of representation levels are reconstruction values that are aligned with the treads of the staircase. ,. What is #eant y id'e !hanne' noise? Idle channel noise is the coding noise measured at the receiver output with ero transmitter input. -.What are the a""'i!ations a""'i!ations of )&ff#an !odin0? a. Image compression b. Audio compression c. Te1t compression d. . What is for*ard and a!9*ard ada"tion? In forward adaption, adaption is done before encoder. ecoder doesnCt 2now about manipulation done. In bac2ward adaption, adaption is done after encoder. here decoder 2nows about manipulation done . What are the a""'i!ations a""'i!ations of M=E56-? igital libraries-This include video libraries, image catalogs, musical dictionaries, future home multimedia databases. 3ultimedia directory services-An e1amples are yellow pages. 0roadcast media selection-This includes radio channel and Internet broadcast search and selection. 3ultimedia editing-personalied electronic news services and media Authoring 144.Des!rie 144.Des!rie the ai#s of M=E5621 To understand understand if and how how various various components components fit together together To discuss which which new standards standards may be re%uired re%uired,, if gaps in the infrastructure e1ist and when the above two points have been reached.To accomplish the integration of different standards. 1, MARKS QUESTIONS 5. iscuss in detail various evaluation techni%ues for compression. 4. )1plain the concept of down sampling and up sampling. 6. )1plain in detail the error analysis and methodologiesW . a )1plain the ta1onomy of compression techni%ue. b )1plain the concept of scalar uantiation theory and rate distortion
theory K. a efine redundancy. &hat are the various types of redundancyW )1plain how redundancy can be removedW b )1plain the following termsD i #ource encoding ii /ector uantiation. . &hat is motion compensation in ".45.briefly discussing itW )1plain model based codingW 9. &rite short notes on video signal representationW :. efine wavelet. iscuss the concept of wavelet based compression techni%ues )1plain with e1amplesW U. iscuss various 3)* standardsW 5B. )1plain the /I technology for symmetric and asymmetric motion video compression$decompositionW 55. )1plain various predictive techni%ues for image compressionW 54. 0ring out difference btw ')* + '0I*W 56. )1plain clearly various processes in ')* image compressionW 5. iscuss contour based compression techni%ue for imageW 5K. )1plain transform codingW 5. )1plain )I?, #I"T, ')*, and '0I*W 59. &rite an algorithm and e1plain the fractal compression techni%ue for images. &hat are its applicationsW 5:. )1plain ?3 with bac2ward adaptive predictionW 5U. &hat is meant by "uffman codingW )1plain its types. *ive some application of "uffman codingW Refer Fhalid #ayoodD Introduction to ata compression, 3organ, "uffman "arcourt, #econd )dition-4BBB. age =oD 6U to 95. 4B. &hat are the applications of arithmetic codingW )1plain anyone application. age =oD 5B to 556. 45. )1plain about Adaptive "uffman coding age =oD KK to B 44. &hat is meant by arithmetic codingW )1plain it with an e1ample. age =oD 9: to ::. 46. efine the concept of predictive techni%ueW The techni%ue techni%ue that uses uses the past values values of a se%uence se%uence to predict predict the current value and then encode the error in prediction or residual is called predictive techni%ue
4. &hat are the predictive techni%ues for image compressionW 5. ?A7I? (conte1t adaptive loss less image compression 4. ')* (Qoint photographic e1perts group 5. 3T (move to front algorithm iscrete walsh hadamard transform 4K. &hat are the various steps of ')* compression processW i. Applying transformation- iscrete cosine transform on pi1els ii. uantiation-scalar iii. )ncoding- "uffman encoding 4. &hat are the various transformation used for image compressionW 4. Farhunen- loeve transform 49.)1plain the concept of linear predictive coder in speech compression 4:.iscuss the role of 3 in sub band coding 4U.)1plain the various audio compression techni%ues 6B.*ive a detailed description of *.944 audio coding scheme 65.)1plain vocoders 64.)1plain an application to speech coding 66.&rite notes on audio-silence compression 6.)1plain the basic sub band coding algorithm 6K.)1plain an application to audio coding.
