Chapter 22
Digital Video Processing Techniques and Applications
What Will We Learn?
- What is motion estimation (ME) and why is it relevant?
- Which techniques and algorithms can be used to estimate motion within a video sequence?
- Which techniques can be used to filter a video sequence?
- What is the role of motion compensation (MC) in video filtering?
22.1 Fundamentals of Motion Estimation and Motion Compensation
Motion is an essential aspect of video sequences. The ability to estimate, analyze, and compensate for relative motion is a common requirement of many video processing, analysis and compression algorithms and techniques. In this section, we present the fundamental concepts associated with motion estimation and compensation.
Motion estimation is the process of analyzing successive frames in a video sequence to identify objects that are in motion. The motion of an object is usually described by a two-dimensional (2D) motion vector (MV), which encodes the length and direction of motion.
Many two-dimensional motion estimation algorithms are based on the concept of optical flow, which can be defined as the pattern of apparent motion of objects, surfaces, and edges in a visual scene. The optical flow—and consequently the apparent motion—can be computed by measuring variations (i.e., gradients) of intensity values over time. The resulting pattern is a vector field—known as the optical flow field—containing the motion vectors for each support region (e.g., pixel, block, or region) ...
