Satellite image segmentation based on YCbCr color space

Abstract

Segmentation is one of the most important processes in the satellite image processing to retrieve most useful information from the satellite images. This paper proposed an effective fuzzy based method of segmentation of satellite images in YCbCr color space. The YCbCr Color space represents color as intensity and exploits the characteristics of human eye. Our eye is more sensitive to intensity than hue. The intensity component can be stored with greater accuracy as the amount of information to be minimized. The JPEG file format mostly uses this color space to discard the unwanted or unimportant information. In the proposed approach, the satellite image in RGB color space is transform into YCbCr color space and then the transformed satellite image is split into three different components (channels or images) based on luminance and chrominance. Subsequently Fuzzy based segmentation is applied separately for all three components for efficient segmentation. Finally the threshold is applied to extract the foreground (object) from the background. The experimental result reveals that the proposed fuzzy based segmentation method is efficient and accurate for extracting the necessary information from the satellite images.