International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) Web Site: www.ijettcs.org Email: [email protected], [email protected] Volume 1, Issue 1, May-June 2012 ISSN 2278-6856 Comparative Study of Motion Estimation & Motion Compensation for Video Compression Jigar Ratnottar1, Rutika Joshi2, Manish Shrivastav3 1RGPV University, Electronics & Communication Engg. Department, Technocrates Institute of Technology, India [email protected] 2 RGPV University, Electronics & Communication Engg. Department, NIIST College, India [email protected] 3 RGPV University, Electronics & Communication Engg. Department, Technocrates Institute of Technology, India [email protected] Abstract: The newest predictive video coding schemes increasingly include the content based techniques which 2. Motion Estimation open new possibilities for advanced compression technology. Respected paper gives the better comparison A video sequence can be considered to be a discredited for Motion Estimation & Motion Compensation, which three-dimensional projection of the real four-dimensional are the major parameter for getting the highest continuous space-time. The real world objects can move, compression. The ME process analyzes previous or rotate, or deform. The movements cannot be observed future frames to identify blocks those are not changed directly, but instead the light reflected from the object and the motion vectors are stored in the place of blocks. surfaces and projected onto an image. The discretization Whereas Motion Compensation technique gives the causes noise into the video sequence, from which the residue of original image & estimated image. video encoder makes its motion estimations. There may Keywords: Motion Estimation, Motion Compensation, Group also be noises in the image capture device (such as a of Pictures (GOP), MPEG. video camera) or in the lines used for electrical transmission. A perfect motion model would take all the factors into account and find the motion that has the 1. INTRODUCTION maximum likelihood from the observed video sequence. The Digital video coding format has gradually increased The movement of objects can change between the frames. since the 90s when the MPEG-1 first emerged. In a model of the object motion between the frames, “the Digitization of video scenes was an inevitable step since it motion that occurred between the reference frame and the has many advantages over analog video. Digital video is current frame” can be estimated by the encoder. This virtually immune to noise, easier to transmit and is able method is known as the motion estimation (ME) [1]. to provide a more interactive interface to users. In the For the better current frame prediction, the encoder used video scene, the data redundancy arises due to changes in the information that move between the reference frame spatial, temporal and statistical correlation between the contents and motion model. This process is called the frames. Due to the difference in their characteristics, the motion compensation (MC), and the prediction so correlation is processed separately. Hybrid video coding produced is called the motion-compensated prediction architectures have been employed since the first (MCP) or the displaced-frame (DF) [2]. The coded generation of the video coding standards, i.e. MPEG. The prediction error signal is known as called the displaced- MPEG format consists three main parts to reduce the data frame difference (DFD). A block diagram of a motion- redundancy from the all three sources described above. compensated coding system is explained in the Figure The Motion estimation and compensation both are used to Generally, The interframe coding is most commonly used reduce the temporal redundancy between the successive method. frames in the time domain. The Transform coding, also The ME use reference frame can occur temporally after or commonly used in the image compression, is employed to before the current frame, it is called forward and reduce the spatial dependency within a frame in the backward prediction, respectively. But in the bidirectional spatial domain. The entropy coding is used to reduce the prediction, two reference frames (one each for forward statistical redundancy over the residue and the and backward prediction) are worked and the two compression data. This technique is also called lossless predictions are interpolated (the resulting predicted frame compression technique; it is commonly used in the file is called B-frame). The block-matching motion estimation compression. (BMME) algorithm is most commonly used ME method. - 33 - International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) Web Site: www.ijettcs.org Email: [email protected], [email protected] Volume 1, Issue 1, May-June 2012 ISSN 2278-6856 preferred because its simple and good compromise between the prediction quality and the motion overhead. There are many other approaches to motion estimation, some using the frequency or wavelet domains, and designers have considered scope to invent new methods since this process does not need to be specified in coding standards. The standards need only specify how the motion vectors should be interpreted by the decoder. Block Matching (BM) is the most common method of motion estimation. Typically each macro block (16×16 pels) in a new frame is compared with the shifted regions of the same size from previous decoded frame and the shift its results in minimum error is selected as the best motion vector for that macro block. After it, motion compensated prediction frame is formed from all the Figure 1. Motion Compensated Video Coding shifted regions of the previous decoded frame [2]. 2.1 Block Matching Algorithm 2.1.1 Backward Motion Estimation Figure 2 illustrates a process of block-matching The motion estimation generally considered as backward algorithm. In the typical Block Matching Algorithm, motion estimation, since the current frame is considered each frame is divided into blocks and each of which as the candidate frame and the reference frame on which consists of the luminance and the chrominance blocks. the motion vectors are searched is a past frame, that is, Usually, for the coding efficiency, the motion estimation the search is backward. Backward motion estimation will performed only on a luminance block. The each leads to forward motion prediction. luminance block in the present frame will matched against candidate blocks in the search area on the reference frame. The candidate blocks are just the displaced versions of the original block. The best candidate block is found and its motion vector is recorded. In the typical interframe coder, the input frame is subtracted from prediction of reference frame. Accordingly, motion vector and resulting error can be transmitted instead of original luminance block; therefore interframe redundancy is removed and data compression is achieved. At receiver side, decoder builds frame difference signal from received data and adds it to reconstructed reference frames. Figure 3. Backward Motion estimation with current frame as k and frame (k-1) as the reference frame 2.1.2 Forward Motion Estimation It is just the opposite of backward motion estimation. Here, the search for motion vectors is carried out on a frame that appears later than the candidates frame in temporal ordering. In other words, the search is “forward”. Forward motion estimation leads to backward motion prediction. It may appear that forward motion estimation is unusual, since one requires future frames to predict the candidate frame. However, this is not unusual, since the candidate frame, for which the motion vector is Figure 2. Motion Estimation being sought is not necessarily the current, that is the most recent frame. It is possible to store more than one This algorithm is use the concept of a translational model frame and use one of the past frames as a candidate frame for the objects motion between frames. Here it is assumes that uses another frame, appearing later in the temporal that, all the pels within a block undergo same order as a reference. translational movement. Apart from it, there are many other the ME methods, but the BMME is normally - 34 - International Journal of Emerging Trends & Technology in Computer Science (IJETTCS) Web Site: www.ijettcs.org Email: [email protected], [email protected] Volume 1, Issue 1, May-June 2012 ISSN 2278-6856 This is also referred to as the optical flow method, which works on the fundamental assumption of brightness constancy that is the intensity of a pixel remains constant, when it is displaced. However, no unique match for a pixel in the reference frame is found in the direction normal to the intensity gradient. It is for this reason that an additional constraint is also introduced in terms of the smoothness of displacement vectors in the neighborhood. The smoothness constraint makes the algorithm interactive and requires excessively large computation time, making it unsuitable for practical and real time implementation. An alternative and faster approach is the block based Figure 4. Forward Motion estimation with current frame motion estimation. In this method, the candidates frame as k and frame (k+1) as the reference frame is divided into non-overlapping blocks ( of size 16 × 16, or 8 × 8 or even 4 × 4 pixels in the recent standards) and Forward motion estimation (or backward motion in the each of such candidate block, best motion vector is compensation) is supported