Chapter 6
Depth image coding using piecewise-linear functions

“One would only need to cut these paintings out... assemble them following the indications given by the colours, to end up in front of a sculpture.”

Pablo Picasso. Spanish painter.

This chapter proposes a novel depth image coding algorithm which concentrates on the special characteristics of depth images: smooth regions delineated by sharp edges. The algorithm models these smooth regions using piecewise-linear functions and sharp edges by a straight line. To define the area of support for each modeling function, we employ a quadtree decomposition that divides the image into blocks of variable size, each block being approximated by one modeling function containing one or two surfaces. The subdivision of the quadtree and the selection of the type of modeling function are optimized, such that a global rate-distortion trade-off is realized. Experimental results show that the described techniques improve the resulting quality of compressed depth images by up to 3 dB, when compared to a JPEG-2000 encoder.

 6.1 Introduction
 6.2 Geometric modeling of depth images
  6.2.1 Overview of the depth image coding algorithm
  6.2.2 Modeling-function definitions
  6.2.3 Estimation of model coefficients
 6.3 Bit-allocation strategy
  6.3.1 Modeling function selection
  6.3.2 Quadtree decomposition
  6.3.3 Quantizer selection
  6.3.4 Summary of the complete algorithm
 6.4 Experimental results
  6.4.1 Coding experiments
  6.4.2 Rendering experiments based on encoded depth
  6.4.3 Comparisons with H.264/MPEG-4 AVC and MVC
 6.5 Conclusions and perspectives