Conditional Quantization for SHVC Quality Scalability of Depth Maps

3D-High Efficiency Video Coding, the latest standard for depth maps compression, is adapted to point-to-point applications where the transmission chain consists of one transmitter and receiver. However, a scalable coding is solicited for multipoint applications where the data source is unique, but serves heterogeneous receivers through different transmission supports. The Scalable High Efficiency Video Coding (SHVC), the latest standard for scalable compression of texture images, can provide the scalability asset for depth maps compression. However, this is at the expense of the depth discontinuities quality. In this paper, we propose a quality scalable compression method adapted to the depth maps context. Besides the enhancement layers that ameliorate the base layer quality in SHVC, we use an importance map to conditionally quantize the transform coefficients of the coding units according to the sub_enhancement layers they include. Our experiments show that our approach allows higher Rate-Distortion trade-offs, compared to the unmodified SHVC.