Type: Process Essays
Sample donated: Andrew Sherman
Last updated: September 12, 2019
guide the method towards the solution. The software eCognitionTM by Definiens33 is an example of an object oriented applications for image segmentation . Thesetype of software must be supported by pre-required information such as contextualor textual one. This is necessary to increase the segmentation result accuraciesand to make them more appropriate for use 34. In Turn this software is not efficientfor different type of multicomponent images. Image segmentation methodscan be either parametric or nonparametric .
There are several reasons which turnsthe parametric statistics unfavorable for specific image segmentation . These can be:1) complexity and inconsistency of the data structures and designs of these imageswhich depends on the technology; 2) defectiveness in the process of image acquisitionsuch as noise, and resolution limitation; 3) The correction tasks (enhancement,filtering, and resolution merge). To fix these limitations are time consuming, thereforeleading to exhaustion of computer and human resources.
In response to theneed for effectiveness in statistical analysis, nonparametric segmentation methodshave been widely used in solving many problems. These methods estimate the underlyingdistributions from the data without any assumptions about the structuresof the distributions. Nonparametric algorithms are applied when the problem of parameterizationis unavailable, without the need to rely on the estimation of parameters.Genetic Algorithm (GA) is a well-known nonparametric algorithm which areused widely in solving many problems including image segmentation . AlthoughGA solves the problems encountered in parametric segmentation methods, it stillneeds to be enhanced to satisfy some critical issues for some applications. Thisissues are high processing speed and at the same time obtaining global optimal solution.
That is why Hybrid GA is used which include heuristic algorithms such asHill Climbing that plays a role in slowing the convergence process toward localoptimal solution and provides the desired global optimal solution. Hybrid GA providesthe desired solution but it is very slow and needs long time to converge. Toincrease the speed of Hybrid GA, variable length chromosomes are used to createDynamic population which reduces the time to perform the reproduction processesand increase the speed of Hybrid GA convergence toward the global solution significantly.Image segmentation techniques and methods can be divided into threecategories supervised which requires complete interference by the user and training.Semi-supervised which requires partial interference such as providing the numberof regions, clusters or classes.
Finally, the unsupervised methods which does notrequire any interference by the user, it is completely automated. Some of the realworldapplications of image segmentation are: Machine vision, Medical imaging.Object detection, Biometrics, Natural resources mapping, Object recognition andmany more.