The Gaussian function (GF) is widely used to explain the behavior or statistical distribution of many natural phenomena as well as industrial processes in different disciplines of engineering and applied science. For example, the GF can be used to model an approximation of the Airy disk in image processing, laser heat source in laser transmission welding [1], practical microscopic applications [2], and fluorescence dispersion in flow cytometric DNA histograms [3]. In applied sciences, the noise that corrupts the signal can be modeled by the Gaussian distribution according to the central limit theorem. Thus, by fitting the GF, the corresponding process/phenomena behavior can be well interpreted.This article introduces a novel fast, accurate, and separable algorithm for estimating the GF parameters to fit observed data points. A simple mathematical trick can be used to calculate the area under the GF in two different ways. Then, by equating these two areas, the GF parameters can be easily obtained from the observed data.
GAUSSIAN FUNCTION FITTING APPROACHESA GF has a symmetrical bell-shape around its center, with a width that smoothly decreases as it moves away from its center on the x-axis. The mathematical form of the GF iswith three shape-controlling parameters, A, µ and σ, where A is the maximum height (amplitude) that can be achieved on the y-axis, µ is the curve-center (mean) on the x-axis, and σ is the standard deviation (SD) which controls the width of the curve along the x-axis. The aim of this article is to present a new method for the accurate estimation of these three parameters. The difficulty of this lies in estimating the three shape-controlling parameters (A, µ and σ) from observations, that are generally noisy, by solving an overdetermined nonlinear system of equations. The standard solutions for fitting the GF parameters from noisy observed data are obtained by one of the following two approaches:1) Solving the problem as a nonlinear system of equations using one of the leastsquares optimization algorithms. This solution employs an iterative procedure such 2 as the Newton-Raphson algorithm [4]. The drawbacks of this approach are the iterative procedure, which may not converge to the true solution, as well as its high cost from the computational complexity perspective. 2) Solving the problem as a linear system of equations based on the fact that the GF is an exponential of a quadratic function. By taking the natural logarithm of the observed data, the problem can be solved in polynomial time as a 3 × 3 linear system of equations. Two traditional algorithms have been proposed in this context: Caruana's algorithm [5] and Guo's algorithm [6]. Furthermore, instead of taking the natural logarithm, the partial derivative is used in Roonizi's algorithm [7].In this article, we will consider only the second approach, which is more suitable for most scientific applications, due to its simplicity and avoidance of the drawbacks of the first approach. Let us start with a brief introduction of the existing three...