Displacement Estimation in Micro-photographies through Genetic Algorithm

The displacement calculation from a pair of images, is a problem without a robust and complete solution. This is due to the several factors that are involved. Particularly in displacement estimation of micro-metric objects on micro-photography, is complicated by dimensions and scales involved. Usually, the process to estimate the displacement comprehends two things: (a) reference zones and similarity criterion of the region of interest in a pair of images [13]. In microphotography images, selecting which regions must be candidates to track is a complicated task due to the level of texture on the image and the light conditions involved. For this reason, normally some error criterion are built, however the numeric method for estimate the displacement no warrant the convergence in a solution that represent physically the observed displacement because they usually fall in local minimum. For this reason, this work presents a proposal based on a Monte Carlo method Index Search, implemented as an evolutionary algorithm [5] which allows to examine the metric-space searching the better solution for the displacement calculation on micro-photographies. Experimentally, a piece of graphite has displaced in controlled increments in the order of millimeters. The results obtained are compared versus the real displacements on the micro-metric table, characterizing the system error.