Interframe prediction
In this article, Dr Ming Yan discusses his recent research.
The purpose of inter-frame prediction is to remove the temporal redundancy of video. In a video sequence, two temporally adjacent frames often have high similarity, so the coded units of the adjacent frames in the temporal domain are generally selected to predict the current coding unit. Motion estimation and motion compensation are two important parts of inter-frame prediction.
Motion estimation
The process of searching for a similar coded unit in the adjacent coded frames of the current frame is called motion estimation. The coded unit is called the reference block, and the frame is called the reference frame. Motion estimation needs to find the best reference block (matching block) and calculate the motion vector (position offset from the current block to the reference block).
Motion compensation
Motion compensation is a method to describe the difference between the current frame and the reference frame, specifically, it describes how the reference block in the reference frame moves to a certain position in the current frame. In general, motion compensation needs to be based on some motion model to calculate the prediction value of the current block.
Process
The specific process of inter-frame prediction coding is to enter the motion estimation process together with the current frame and the reference frame (that is, the previous reconstructed frame output by the decoder), and obtain the best matching block and motion vector MV through motion estimation. The motion vector MV is input into the motion compensation process, which uses the motion vector MV to compensate the motion of the input reconstructed reference frame. In the process of motion compensation, the predicted pixel value and the difference between it and the original pixel value, namely the residual, are obtained. After receiving the residual bitstream signal, the decoder performs inverse quantization and inverse transform, and calculates the reconstructed pixel value of the current encoded frame together with the predicted pixel value obtained after motion compensation.
Your task
Why do you think inter-frame prediction is necessary?
Share your thoughts and ideas in the comments below.
Reach your personal and professional goals
Unlock access to hundreds of expert online courses and degrees from top universities and educators to gain accredited qualifications and professional CV-building certificates.
Join over 18 million learners to launch, switch or build upon your career, all at your own pace, across a wide range of topic areas.
