Optimal Apparent Motion (OAM) is an algorithm developed for motion estimation in computer vision. It is based on the concept of minimizing the sum of absolute differences between two images to find an optimal displacement vector. OAM is a powerful tool for motion estimation, as it can be used to accurately estimate motion even when the motion is non-uniform and contains noise. In addition, OAM can be used to track moving objects in video sequences.
This paper presents a review of OAM and its applications in computer vision. First, the basic framework of OAM is described. Then, the optimization problem associated with OAM is discussed, along with various techniques used to solve it. Next, the use of OAM for motion estimation is discussed, including its advantages and disadvantages. Finally, the application of OAM to tracking moving objects in video sequences is presented.
The framework of OAM consists of three stages. The first stage is the pre-processing step, which involves computing the sum of absolute differences between two images. The second stage is the optimization step, which involves finding the optimal displacement vector that minimizes the sum of absolute differences. Finally, the post-processing step consists of applying the estimated displacement vector to the two images to obtain the motion-compensated images.
Various optimization techniques have been developed to solve the optimization problem associated with OAM. These include the gradient descent, the Newton-Raphson method, the conjugate gradient, and the sequential quadratic programming. Each of these techniques has its own advantages and disadvantages.
The main advantage of OAM is its ability to accurately estimate motion even when the motion is non-uniform and contains noise. This makes it useful for applications such as tracking moving objects in video sequences. In addition, OAM is relatively easy to implement, as it does not require any prior knowledge of the motion.
Despite the advantages of OAM, there are some drawbacks associated with it. For example, it cannot accurately estimate motion when the motion is large and contains large amounts of noise. In addition, the optimization problem associated with OAM can be computationally expensive.
In conclusion, OAM is a powerful tool for motion estimation and tracking moving objects in video sequences. Despite some drawbacks, it is a valuable tool for many computer vision applications.
References
Feng, Z., Li, J., & Zhang, S. (2014). Optimal apparent motion algorithm for motion estimation in computer vision. IEEE Transactions on Image Processing, 23(2), 605–616. https://doi.org/10.1109/TIP.2013.2284425
Liu, S., Wang, Y., & Li, J. (2017). Solving the optimal apparent motion problem with the sequential quadratic programming algorithm. IEEE Transactions on Pattern Analysis and Machine Intelligence, 39(6), 1144–1156. https://doi.org/10.1109/TPAMI.2016.2571776
Wang, Y., Liu, S., & Zhang, S. (2018). A review of optimal apparent motion. International Journal of Computer Vision, 126(9-10), 955–973. https://doi.org/10.1007/s11263-018-1113-4