Modeling of DCT Coefficients in the H.264 Video Encoder-Edición Única

DCT stands for Discrete Cosine Transform (Ahmed, Natarajan, & Rao, 1974); it is an important tool used in video and image compression systems (Richardson, 2003). The understanding of the DCT coefficients’ statistical distribution has many applications; one of them is the rate control for video coding, which requires knowledge of the rate-distortion relation as a function of the encoder parameters and the video source statistics (Kamaci & Altunbasak, 2005) This thesis focuses in the rate and distortion relations as a function of the quantization parameter and the statistics of the DCT coefficients. The objective of this work is to compare three theoretical Probability Density Functions (Laplace, Cauchy and Generalized Gaussian) in terms of: how well they represent the empirical DCT coefficient data; and how well the models based on them can predict the actual distortion and the actual rate of the coded video.