Rate-complexity-distortion optimization for quadtree-based DCT coding

Abstract

We study rate-complexity-distortion (R-C-D) tradeoffs for video coding, where we focus on the complexity of computing the inverse DCT. A quadtree coding approach is used and the quadtree is optimized based on the constraints of not only a rate budget but also decoding complexity budget. We employ a variable complexity algorithm (VCA) for the IDCT in order to obtain better complexity result. The main novelty of this work is to demonstrate that VCA approaches combined with R-C-D optimization can provide better results (e.g., lower rate at same distortion and complexity) than approaches proposed in the past, which relied on fixed blocksize and fixed complexity techniques.