Motion Compensation

Under the heading motion compensation or motion Prediction ( literally: motion prediction ) is a set of algorithms summarized, which are used mainly supportive in video compression. The temporary ratios between adjacent frames are used to minimize the total amount of data to be compressed. Simplified, this means that image areas that look over multiple frames of time very similar, not a second time be saved completely, but only the change in their position relative to other images.

Simple compression techniques reduce only immobile parts of the image, for example, the background against which a newscaster sitting, or simple camera motion such as zoom, horizontal or vertical shift. However, current video codecs use more complex algorithms that can compress and moving parts of the image accordingly. Here, the direction of movement of an object is predicted and described by a motion or displacement vector. From then on, only the corresponding displacement vector must be stored, which significantly less memory than the actual image data, which leads to some enormous space savings.

In the end, the first image of the scene is initially stored only once. Then only those parts of the image must be re- filed for the following frames, which were originally occupied by the moving object. However, usually the entire image at regular intervals is re- stored to both increase the fault tolerance when (not coils) to allow the back and forth jumping within a video.

Block Motion Compensation

Block motion compensation (BMC ) is used among other things in the MPEG- 2 standard. In this method, each frame is divided into n by n pixel large blocks which are compared with the previous frame. Found there a similar picture, so only the displacement vector is stored at the moves this snippet.

A block of pixels can also be interpreted as a matrix. Now let Bx, y ∈ Nn × n pixel block of the current frame with the horizontal offset n ⋅ x and the vertical offset n ⋅ y. The matrix elements Bi, j are the pixel at the position (s ⋅ x i, y n ⋅ j), Pi, j corresponding to the pixels in the previous frame. In order to find the displacement vector is compared Bx, y with the vector (u, v) shifted image in the previous frame. An indicator of the difference you, v between pixel block and framing is the sum of squared differences ( Sum of Squared Differences, SSD):

Other indicators could be, for example, the sum of absolute differences ( Sum of Absolute Differences, SAD) or the cross-correlation. Man you calculated v for all shifts within a specified proximity of Bx, y and selects the vector ( u, v) as the displacement vector in which you, v ≤ dmax and is the minimum of all differences. If, dmax = 0, then the original data stream is fully reconstructed in a dmax > 0, the motion is no longer compressed lossless Compensation.

An improvement of the BMC algorithm can be achieved by addition to the previous and the subsequent frame is used to determine the displacement vector. Meanwhile, motion compensation algorithms to further Motion Prediction were developed based on the block, which provide in some cases significantly better results. Firstly, here is the Variable Block-Size Motion Compensation Overlapped Block Motion and called on the other hand Compensation.

Quarter Pixel Motion Compensation

Quarter Pixel Motion Compensation (often only briefly, QPel, quarter pixel or Q- Pel called ) is compensation with motion vectors of the accuracy of a quarter pixel in digital video encoding motion. Qpel is a better prediction of movements and thus produces better image quality and small file sizes.

Quarter Pixel Motion Compensation is part of the MPEG-4 ASP (optional) and AVC standard ( mandatory ) and is often used in conjunction with Global Motion Compensation (only ASP ) was used.

Global Motion Compensation (GMC )

Global Motion Compensation (GMC ) is used for digital video coding, particularly MPEG-4 ( for example Xvid). In contrast to block-based motion compensation motion between two frames is described with a global affine transformation. This also movements that go beyond the mere shifting, like rotations or distortions can be described. GMC's aim is a better encoding quality for fast-moving scenes, such as explosions. The use of GMC results in better image quality and smaller file size, but which is associated with a higher coding effort. GMC is a relatively new process that is not yet supported by all players and thus can lead to problems.