Automated land resource classification of electronic photograph based on satellite CMOS detector

The research of automated classifying and recognizing electronic photograph based on satellite detectors with various machine learning methods has been in the limelight of current research. In our experiment, we applied K-means clustering algorithm to segment urban electronic photograph of CMOS detector, and interpreted segmentation results as different object classes such as greenbelt and sandy desertification, serving for machine intelligence application of remote electronic sensor equipments. In the paper, first we discussed the difference between traditional segmentation methods and K-means segmentation methods, and then compared gray image segmentation with color based image segmentation; furthermore we did proper semantic analysis research on the object classes of segmentation results and made relevant statistics. Experiment results show that, K-means clustering can get preferable results when electronic photograph has high discrimination in color scales L*a*b*values. Statistical analysis of segmentation results has great scientific significance and practical value for both city planning and automated resource monitoring.