Method and apparatus for providing complexity balanced entropy coding

What is this patent about?

’701 is related to the field of video coding and decoding, specifically addressing the computational bottlenecks associated with entropy coding in high-bitrate environments. In modern video standards like H.264/AVC, advanced entropy coding techniques provide high compression efficiency but demand significant processing power, which can drain battery life and overwhelm the hardware of mobile devices. The invention seeks to balance this trade-off by optimizing how different types of data within a video stream are processed.

The underlying idea behind ’701 is that not all video data requires the same level of complex processing to achieve efficient compression. By identifying which data elements appear most frequently—such as residual transform coefficients—the system can apply a simplified coding path for the high-volume data while reserving computationally intensive context adaptive updates for less frequent elements. This selective approach prevents the processing load from scaling linearly with the bitrate, effectively capping the complexity without sacrificing overall performance.

The claims of ’701 focus on a method and apparatus that categorize syntax elements into two distinct groups based on a frequency of occurrence threshold. The first category consists of elements that appear relatively infrequently and are processed using entropy coding that includes a context update stage to maximize efficiency. The second category includes high-frequency elements that bypass the context update stage entirely, instead using a simplified entropy coding process with fixed or targeted probabilities.

In practice, the invention works by binarizing high-frequency symbols, such as DCT coefficients, to target a fifty-percent probability for each bin. By treating these bins as having equal likelihood, the encoder can utilize a bypass coding mode within an arithmetic coding engine, skipping the most taxing part of the process: the constant recalculation of probability models. This ensures that even as the video quality and bitrate increase, the number of complex context updates remains manageable for the hardware.

This approach differs from prior solutions by moving away from a rigid, profile-based choice between low-complexity and high-efficiency coding. Instead of forcing a device to choose between a simple but inefficient coder and a powerful but demanding one, ’701 implements a unified entropy coding design. By signaling the categorization at the slice or picture level, the invention allows a single codec to dynamically adapt its workload, maintaining a complexity-balanced output that is optimized for the specific constraints of mobile hardware.