Precise and effective segmentation of satellite images is a crucial step in the field of remote sensing. A suitable segmentation technique may accurately depict distinct objects in an image using a reduced number of colour levels while preserving significant information about the objects. This is the most challenging task to find an optimum number of colour levels. The work used the genetic algorithm (GA), which draws inspiration from natural selection, to identify the optimal set of threshold values for achieving the highest quality of segmentation. The approach utilizes parallel processing to apply the genetic algorithm to each channel of the satellite image and execute multilevel segmentation in parallel on the three dedicated cores of the system. Finally, the combine operation is performed on the optimal threshold values for each channel to generate the colour-segmented output. Due to its parallel execution, it requires less time compared to sequential execution, resulting in faster segmentation. The experiment was conducted using sample images of Landsat-8. Various parametric analyses have been conducted on the segmented images, and the results indicate that the approach is sufficiently suitable for colour satellite image segmentation.