Optimization of the Histogram Intervals Number which Approximate Brightness Probability Distributions in Stochastic Image Alignment Based on Information Similarity Measures

This article investigates the influence of histogram intervals number on the accuracy and speed of the stochastic algorithm needed to solve the image alignment problem. The mutual information of images is considered as a similarity measure. Histograms are used for approximation of brightness probability density distributions when finding the gradient of image mutual information. Experimental results are obtained for the Shannon, Renyi and Tsallis mutual information cases.

I. Introduction

The image alignment procedure is demanded at the decision of applied problems of image processing in various areas [1]–[3]. In particular, for the analysis of satellite images, it is important to provide accurate georeferencing or metric correctness.