Introduction to 2D Target Tracking Using Kalman Filter ECE Project:
For the motion estimation recursive approaches has become a common method. It helps in finding the Kalman filtering techniques both in the respect of frequency and time domain. This approach is based on a block based technology and usually this kind of technology are not widely available. This technology helps in refining the motion vector estimation that is derived from fast algorithms. Again, this paper proposes an estimation based on the object motion that make use of the Kalman filtering technique in order to modify the estimates of motion that is derived from the Kalman application and the three step algorithm.
The results that are presented above are quite interesting and encouraging as well from the aspect that with appropriate model and priori assumptions that are close to vector and real motion behavior we can easily go through the Kalman filtering creates a greater PSNR in comparison to other techniques of any sequence. It is proved that simple formulations are the base of the Kalman filter and this paper gives more suitable options for class A sequence of images. From this point few questions can arise in the mind of the reader like:
Shall we consider the accuracy of sub-pixel?
Shall we make the state-space representation more elaborate for the extended Kalman filter motion?
Shall we leave the components of motions on independent assumptions to consider the correlation that usually exists between the x ans y direction displacement?
All the questions that are stated above are completely feasible and keeping in mind the usefulness of the implementation of the approaches this become a trade-off matter among the estimates of the resulting motion, the real visual image quality and computational complexity. This method is very helpful and if handled in the proper manner and direction then it is possible to get the accurate results.