Survey of Particle Swarm Optimization and Random Forest based Land Cover
Forest ClassifierØ This algorithm could
segregate two landscapes with 90% and 100% accuracy.Ø Machine
Learning. Ø DATA SET1.
data of Spain.2.
images of Canadian Prairies.3.
and predictor variable data extracted from GIS database. ·
Swarm OptimizationØ This algorithm
starts with a population of randomly selected particles moving through the
search space with a velocity that is automatically tuned with respect to its
behavior and other particles in the population.Ø Helpful in
depicting the aggregated behavior of dispersed, self-organized structures (E.g.
Ants, birds flock).Ø Artificial Intelligence. Ø DATA SET4. The data set in
the form intensity value are taken from ERDAS.5. ERDAS, Matlab,
Rosetta on a multi-spectral image of Alwar area in Rajasthan.6.
Earth imagery. AdvantagesHigh accuracy and efficiency than other techniques.Based on the a variety of data set used the result
depicts clearly that these techniques are the better techniques to classify
images based on example data set(training data set).Even the comparison of results with the results of
other algorithms show that these two techniques are better. DisadvantagesBoth techniques use data sets to train the system
first and then work on new data set.This data set include same area in different
weather, seasonal conditions.Training data set acquisition requires more time.PSO requires dataset acquired in a short interval of
time. Usable in our projectBoth techniques deals with classification of
landscape images. Since they deliver more accurate results, they may be used to
classify Satellite Images. Remote Sensing Image ClassificationDetection by
Classification of Buildings in Multispectral Satellite Imagery