Uni-modal based fast half-pel and fast quarter-pel refinement for video encoding

Abstract

A method of half-pixel interpolation and quarter-pixel interpolation are adapted for reducing the impact of aliasing within motion estimation. To estimate a motion vector with quarter-pixel accuracy more efficiently, the improved method is able to skip checking certain points using the uni-modal assumption. In an embodiment, a diamond based refinement is implemented. Within the diamond based refinement are half-pel refinement and quarter-pel refinements. Furthermore, within the half-pel refinement are methods for on-the-fly interpolation and pre-computed interpolation. Within quarter-pel refinement, the method depends on whether four neighbor half-pel points are checked or just one or two half-pel points. Moreover, within each of the different embodiments is the ability to focus on quality or speed wherein different methods are implemented to maximize the desired function. In another embodiment, a square based refinement is implemented.

Description

FIELD OF THE INVENTION[0001]The present invention relates to the field of video compression. More specifically, the present invention relates to improved motion estimation in digital video encoders.BACKGROUND OF THE INVENTION[0002]A video sequence consists of a number of pictures, usually called frames. Subsequent frames are very similar, thus containing a lot of redundancy from one frame to the next. Before being efficiently transmitted over a channel or stored in memory, video data is compressed to conserve both bandwidth and memory. The goal is to remove the redundancy to gain better compression ratios. A first video compression approach is to subtract a reference frame from a given frame to generate a relative difference. A compressed frame contains less information than the reference frame. The relative difference can be encoded at a lower bit-rate with the same quality. The decoder reconstructs the original frame by adding the relative difference to the reference frame.[0003]A...

AI Technical Summary

Benefits of technology

[0015]A method of half-pixel interpolation and quarter-pixel interpolation are adapted for reducing the impact of aliasing within motion estimation. To estimate a motion vector with quarter-pixel accuracy more efficiently, the improved method is able to skip checking certain points using the uni-modal assumption. In an embodiment, a diamond based refinement is implemented. Within the diamond based refinement are half-pel refinement and quarter-pel refinements. Furthermore, within the half-pel refinement are on-the-fly interpolation and pre-computed interpolation. Within quarter-pel refinement, the method depends on whether four neighbor half-pel points are checked or just one or two half-pel points. Moreover, within each of the different embodiments is the ability to focus on quality or speed wherein different methods are implemented to maximize the desired function. In another embodiment, a square based refinement is implemented.

Problems solved by technology

Furthermore, it was desired to make these improvements without such a large increase in complexity that the design is impractical to implement.

However, there is a trade-off between the size to which a frame can be compressed versus the processing time and resources required to encode such a compressed frame.

The ratio of I, P and B-frames in the GOP structure is determined by the nature of the video stream and the bandwidth constraints on the output stream, although encoding time may also be an issue.

This is particularly true in live transmission and in real-time environments with limited computing resources, as a stream containing many B-frames can take much longer to encode than an I-frame-only file.

One of the most time-consuming components within the encoding process is motion estimation.

Motion estimation-related aliasing is not able to be avoided by using inter-pixel motion estimation, and the aliasing deteriorates the prediction efficiency.

As a consequence, the computation complexity of searching the half-pixel motion vector and quarter-pixel motion vector becomes dominant.

Images