Statistical Inference on Random Fields in the IEEE Proceedings while looking for a topic for my Ph.D. dissertation. Methods for estimating parameters and testing hypotheses in two-dimensional non-causal models were the subject of this article. This paper led me to classic papers as well as a paper on parameter estimation in Gaussian Markov random field models. I was naturally drawn to the possibilities of using a mathematical statistical framework for computer vision problems because I had been exposed to fundamental concepts in parameter estimation, random processes, and decision theory as a student of electrical and computer engineering. My dissertation dealt with stochastic models for understanding and processing images. Since then, I have worked on computer vision problem-solving strategies based on mathematical statistics. Mathematical statistics tools are very helpful in solving computer vision problems because the majority of them involve inferring some properties (radiometric, geometric, etc.) from images and videos. For mathematical statisticians, computer vision problems can be extremely challenging when it comes to inferring 3D geometry from images and videos. The ability to use appropriate distributions to account for degradations in the data is another reason why statistical methods may be useful for computer vision issues; A Bayesian framework can also take into account any previous data. Manifolds, non-parametric inference tools, and other tools, it's possible to have even more fun.
HTML PDFShare this article
Journal of Applied & Computational Mathematics received 1282 citations as per Google Scholar report
