INTRODUCTION
With the advancements in the multimedia and VLSI technologies, there has been a drastic increase in
the demand and the variety of applications related to video communications and streaming in the last fifteen
years. However, with the increase in applications and features offered to the users, the demand for bandwidth
has been increasing at a very fast pace. Hence, new techniques are being developed to efficiently use the
bandwidth while maintaining a standard in transmitting video data.
One of the methods is to model the system (video) and generate the parameters. These parameters, along with
other required information, are sent over the channel. There are various ways to model a system and obtain the
required parameters from it. Motion Estimation is an important process in getting the required parameters of the
model. However, at the receiver’s end, reconstructing the video or its frames from the received parameters may
lead to errors. Although the errors cannot be eradicated completely, they can be minimized.
There are many methods to minimize the errors while coding. One of the earlier and popular algorithms is the
Exhaustive Block Matching Algorithm (EBMA). In this algorithm, there is an anchor frame and a target frame.
The motion vectors in the target frame are found with respect to the anchor frame. There is a search range box
(-R to R = 2R+1 pixels) which moves around the target frame and is used to find the motion vectors. The box is
moved by 1 pixel every iteration in the horizontal and vertical directions until it reaches the end of the target
image.
There are two types of EBMA algorithm – integer-pel and fractional-pel algorithms. In this project, we
discuss the integer-pel and half-pel algorithms.
