In this paper, a method for designing low-power video schemes is presented. Algorithms that imply a very low dissipation are required for new applications where the energy source is limited, e.g. mobile phones including a camera and video features. Whereas it can be observed that video standards are mainly designed around coding efficiency, we propose to take into account power consumption characteristics directly when designing the algorithm. More precisely, we give some guidelines for the design of low-power video codecs in the scope of modern hardware architectures and we introduce the notion of power scalability. We present an original encoder based on so-called 'Collocated Motion Estimation' designed using the proposed methodology. Experimental results show that we remain close to the coding efficiency of the reference H.264 baseline encoder while the power consumption is largely reduced in our solution. Moreoever this encoder is scalable in memory transfer and computational complexity.
In this paper, we present a novel H.264 video decoder where memory transfers and energy consumption are significantly reduced. A power-friendly solution is indeed required in mobile applications, where autonomy is a key feature. Whereas low-power design is usually regarded as a pure architectural and implementation problem, our solution is based on an algorithmic approach. We first observe that in modern hardware architectures, computational complexity plays a second role in terms of energy dissipation, which is dominated by memory transfers. We thus identify the motion compensation module as the most consuming part of the standard H.264 decoder, due to its numerous memory accesses to the reference frame(s). Through an algorithmic modification that uses concurrent embedded compression techniques and relies on the data structure of the memory, we reduce memory transfers, and hence power dissipation, while maintaining the visual quality. Experimental results prove that the new method is close to the reference H.264 baseline decoder both in terms of objective and subjective measurements. In the meantime, memory transfers have been reduced by 55% in average, implying power savings which lengthen the battery life of the mobile device, increase the reliability of the chip and lower production costs.
Today, 3D subband (3DS) video coding schemes are close to current standard solutions in terms of coding efficiency while they add the scalability functionality through embedded bitstreams. Spatial scalability may be regarded as a key-feature brought by such codecs, enabling adaptation to varying terminal capabilities and display sizes. However, this functionality still suffers from a lack of coding efficiency when motion compensation is used. Though enabling motion compensation at the temporal filtering stage dramatically improves energy compaction, it generates a strong motion vector overhead and introduces a reconstruction drift when decoding at lower spatial resolutions. This is because the Discrete Wavelet Transform and the motion compensation (MC) are not commutative. Some authors solve this problem by transmitting the drift signal as side information, but this increases the bit-rate. In this paper we present a low-resolution optimized MC-3DS scalable codec with no drift, that does not generate any overhead. Its structure is fully compliant with any subband-tree entropy encoder and preserves all the other scalability functions (temporal and SNR). Tests and simulations show that the new scheme remains quite efficient when decoding at full resolution, while it outperforms the previous solution as far as spatial scalability is concerned, especially with high-activity sequences.