英语论文网

JSCC for 3D video transmission application is proposed in [[xi]]. Unlike 2D video, two data streams, left and right views, needs to be considered for source coding of 3D content. This study considers the operation between H.264/AVC and rate compatible punctured turbo codes (RCPT) for source and channel coding respectively. To protect compressed video data from channel errors, the concept of unequal error protection (UEP) is employed to assign different levels of protection to each encoded data partition with regard to their decoding importance. The JSCC method introduced in this chapter aims to improve the performance of 3D video based on the colour-plus-depth 3D representation over the WiMAX based Reyleigh fading channel from a perceptual quality point of view.

1.2经典率失真理论——1.2 Classical Rate-Distortion Theory

Rate distortion theory is a key aspect of information theory which provides the theoretical bounds for lossy data compression. Theories of rate distortion were created by Claude Elwood Shannon, known as “father of information theory”, in 1948 [[xii]], in his initial work on information theory. This addresses the problem of determining the minimal amount of information or entropy (R) that should be communicated over a channel, such that the source (input signal) can be approximately reconstructed at the receiver (output signal) without exceeding a given distortion (D). The relationship between rate and distortion is illustrated in 5‑1. Term rate is usually refers to the number of bits per data sample to be transmitted or stored and “distortion” refers to the degree of difference between original and reconstructed signals, usually evaluated by the mean squared error (MSE). However, as most lossy compression techniques operate on data samples that will be ultimately perceived by human consumers (e.g. watching video and pictures) the distortion measures should be intelligent to model on human perception. To date, since the human perception models are less well developed for image and video, lossy compression techniques still rely on simple statistical measure such as MSE, although with less correlation with regards to HVS.

Ideal noiseless transmission channels should produce the same massages or symbols, X, emitted by the source at the destination. However transmission impairments, such as noise, in the channel alter the emitted symbols, resulting in a different symbol space Y at the receiver. Consider a simple model of an error pone communicational channel shown in xxx. Let's assume XXXX illustrates the forward transitions of the channel. Here X and Y represent the alphabets of symbols transmitted and received respectively, during a unit time over the channel. Let, P(yj|xi) defines the conditional probability distribution functions of output symbols yj for a given input xi, where xiX and yj Y.

If the channel is intended to deliver y-j when xi transmitted, then the error probabilities are defined by P (yj|xi) for all j ≠ i. For the channel, mutual information, I(x-i;yj), measures the amount of information that symbols xi and yj convey about each other. I(x-i;yj), is defined as follows:

(5.1)
In practice, most transmission channels are consider to lie between perfect transfer (i.e. each yj uniquely identifies a particular xi) and zero transfer (i.e. yj is totally unrelated to xi ) Average mutual information is def本论文由英语论文网提供整理，提供论文代写，英语论文代写，代写论文，代写英语论文，代写留学生论文，代写英文论文，留学生论文代写相关核心关键词搜索。