The video compression have many applications and one of the application is Intelligent Transportation Systems (ITS) are also using compressed videos to drive autonomous vehiclesMotion estimation and compensation (ME/MC) are most computationally extensive components of video codecsThe main idea of the project is to implement ME/MC on hardware description language (HDL).The main motivation for us to do this project is its vast use and relevance with future technology
Objectives
Implementation of motion estimation for FPGA
Providing fast processing solution
Understanding the basic use and handling of FPGA.
Understanding and implementing hardware description language (HDL).
Socio-Economic Benefit
NVIDIA - They are working on developing advanced computer vision algorithms for motion estimation in autonomous vehicles, robotics, and other applications.
Intel - They are developing advanced algorithms and hardware for real-time motion estimation in video streams.
Apple - They are working on motion estimation algorithms for their devices such as the iPhone, iPad, and Apple Watch.
Google - They are developing motion estimation algorithms for various applications such as video stabilization, object tracking, and augmented reality.
Sony - They are working on motion estimation algorithms for their cameras and other imaging devices.
Samsung - They are developing motion estimation algorithms for their mobile devices and other applications.
Methodologies
Block Matching Algorithm for Motion Estimation:
The foremost point about block matching motion estimation is that there is high correlation between each pixel and its neighbors. Therefore, assigning a motion vector to a block of pixels is more useful than to an individual pixel. In the block matching motion estimation process, a frame is segmented into n × n blocks. Then, for the largest motion displacement of ‘p’ pixels per frame, the current block is matched with a corresponding block in the previous frame in same coordinates. The best match on base in matching principle gives the displacement.
Why choosing Block Matching Algorithm for Motion Estimation:
There are number of algorithms through which motion estimation can be achieved but we have selected Block Matching Algorithm for our project because it is most widely used and it yields the best results as compared to other algorithm available for motion estimation because this algorithm gives the highest PSNR. Although the complexity of this algorithm is higher but the output is much cleaner and have very less noise.
Outcome
The final outcome of this project is a fully functional and efficient motion estimation module implemented on an FPGA using Verilog. The project contributes to the field of video processing by providing a hardware solution that can be integrated into various video applications. The implemented module can serve as a foundation for future research and development in real-time video processing systems
