A Novel Computer Vision Approach for Segmentation and Classification of Satellite Image using K-Means Algorithm in comparison with Random Forest to Improve Precision

A real time remote sensing satellite image segmentation and classification using machine learning algorithm of K-Mean algorithm and comparison with the Random Forest to increase the precision rate. 2057 images of areas are taken as samples are taken to find the better precision and sensitivity of the satellite image segmentation and classification. The classification and segmentation of satellite images is carried out using the K-Means algorithm where the number of samples K-Means (N=11) and Random Forest (N=11) with the total number of sample size (N=22) The precision for the proposed system K-Means is 96.97% when compared to Random Forest with 93.65%. There is a significant difference in precision (P=0.002) and sensitivity (P=0.299). By comparing the K-Mean algorithm and Random Forest algorithm the precision and sensitivity is high and significant in the K-Mean algorithm.