- Rate–distortion theory
Rate–distortion theory is a major branch of
information theorywhich provides the theoretical foundations for lossy data compression; it addresses the problem of determining the minimal amount of entropy(or information) "R" that should be communicated over a channel, so that the source (input signal) can be approximately reconstructed at the receiver (output signal) without exceeding a given distortion "D".
Rate–distortion theory gives theoretical bounds for how much compression can be achieved using lossy compression methods. Many of the existing audio, speech, image, and video compression techniques have transforms, quantization, and bit-rate allocation procedures that capitalize on the general shape of rate–distortion functions.
Rate–distortion theory was created by
Claude Shannonin his foundational work on information theory.
In rate–distortion theory, the "rate" is usually understood as the number of
bits per data sample to be stored or transmitted. The notion of "distortion" is a subject of on-going discussion. In the most simple case (which is actually used in most cases), the distortion is defined as the variance of the difference between inputand outputsignal (i.e., the mean squared errorof the difference). However, since we know that most lossy compressiontechniques operate on data that will be perceived by human consumers (listening to music, watching pictures and video) the distortion measure should preferably be modeled on human perceptionand perhaps aesthetics: much like the use of probabilityin lossless compression, distortion measures can ultimately be identified with loss functions as used in Bayesian estimation and decision theory. In audio compression perceptual models, and therefore perceptual distortion measures, are relatively well developed and routinely used in compression techniques such as MP3or Vorbis, but are often not easy to include in rate–distortion theory. In image and video compression, the human perception models are less well developed and inclusion is mostly limited to the JPEGand MPEGweighting (quantization, normalization) matrix.
The functions that relate the rate and distortion are found as the solution of the following minimization problem:
Here "Q""Y" | "X"("y" | "x"), sometimes called a test channel, is the conditional
probability density function(PDF) of the communication channel output (compressed signal) "Y" for a given input (original signal) "X", and "I""Q"("Y" ; "X") is the mutual informationbetween "Y" and "X" defined as
where "H"("Y") and "H"("Y" | "X") are the entropy of the output signal "Y" and the
conditional entropyof the output signal given the input signal, respectively:
The problem can also be formulated as Distortion-Rate function, where we find the supremum over achievable distortions for given rate constraint. The relevant expression is:
The two formulations lead to functions which are inverses of each other.
The mutual information can be understood as a measure for "prior" uncertainty the receiver has about the sender's signal ("H(Y)"), diminished by the uncertainty that is left after receiving information about the sender's signal ("H"("Y" | "X")). Of course the decrease in uncertainty is due to the communicated amount of information, which is "I"("Y"; "X").
As an example, in case there is "no" communication at all, then "H"("Y" |"X") = "H"("Y") and "I"("Y"; "X") = 0. Alternatively, if the communication channel is perfect and the received signal "Y" is identical to the signal "X" at the sender, then "H"("Y" | "X") = 0 and "I"("Y"; "X") = "H"("Y") = "H"("X").
In the definition of the rate–distortion function, "D"Q and "D"* are the distortion between "X" and "Y" for a given "Q""Y" | "X"("y" | "x") and the prescribed maximum distortion, respectively. When we use the
mean squared erroras distortion measure, we have (for amplitude-continuous signals):
As the above equations show, calculating a rate–distortion function requires the stochastic description of the input "X" in terms of the PDF "P""X"("x"), and then aims at finding the conditional PDF "Q""Y" | "X"("y" | "x") that minimize rate for a given distortion "D"*. These definitions can be formulated measure-theoretically to account for discrete and mixed random variables as well.
An analytical solution to this
minimization problemis often difficult to obtain except in some instances for which we next offer two of the best known examples. The rate–distortion function of any source is known to obey several fundamental properties, the most important ones being that it is a continuous, monotonically decreasingconvex (U) function and thus the shape for the function in the examples is typical (even measured rate–distortion functions in real life tend to have very similar forms).
analyticalsolutions to this problem are scarce, there are upper and lower bounds to these functions including the famous Shannon lower bound(SLB), which in the case of squared error and memoryless sources, states that for arbitrary sources with finite differential entropy,
where "h(D)" is the entropy of a Gaussian random variable with variance D. This lower bound is extensible to sources with memory and other distortion measures. One important feature of the SLB is that it is asymptotically tight in the high distortion regime for a wide class of sources and in some occasions, it actually coincides with the rate–distortion function. Shannon Lower Bounds can generally be found if the distortion between any two numbers can be expressed as a function of the difference between the value of these two numbers.
The Blahut–Arimoto algorithm is an elegant iterative technique for numerically obtaining rate–distortion functions of arbitrary finite input/output alphabet sources and much work has been done to extend it to more general problem instances.
Memoryless (independent) Gaussian source
If we assume that "P""X"("x") is Gaussian with
varianceσ2, and if we assume that successive samples of the signal "X" are stochastically independent(or, if your like, the source is "memoryless", or the signal is "uncorrelated"), we find the following analytical expressionfor the rate–distortion function:
The following figure shows what this function looks like:
Rate–distortion theory tell us that "no compression system exists that performs outside the gray area". The closer a practical compression system is to the red (lower) bound, the better it performs. As a general rule, this bound can only be attained by increasing the coding block length parameter. Nevertheless, even at unit blocklengths one can often find good (scalar) quantizers that operate at distances from the rate–distortion function that are practically relevant.
This rate–distortion function holds only for Gaussian memoryless sources. It is known that the Gaussian source is the most "difficult" source to encode: for a given mean square error, it requires the greatest number of bits. The performance of a practical compression system working on—say—images, may well be below the "R(D)" lower bound shown.
* [http://www-ict.its.tudelft.nl/vcdemo VcDemo Image and Video Compression Learning Tool]
Wikimedia Foundation. 2010.