Color Features Based Model for Land Cover Identification and Agriculture Monitoring with Satellite Images

Abstract

Color plays a vital role in our visual world, as tones of color are more attractive than shades of gray. It has a significant impact on image analysis, which could be portrayed and used in various forms. For specifying a color in some standard form, color space models are used. Research in the past has revealed that different color spaces have a different perceptual power for identifying patterns, depending on which color space model is relevant for a particular application. The color model literature reveals the nature of varying color space models focused on spectral quantifications, human vision, physiological motivation, and system orientation. In the area of satellite imagery, various color spectral bands in the visible spectrum have been utilized for image acquisition and land cover monitoring, but very less work has been proclaimed using color space models for satellite data based land cover monitoring. The land cover maps obtained using satellite data play a significant role in distinct applications, including agricultural landscape monitoring. This study presents a color space model-based algorithm useful for land cover mapping and agriculture monitoring. The technique was based on the behavior shown by different land cover classes in each of the color spaces. The algorithm is adaptive, as it does not require any prior knowledge and is independent of individual pixel values of the bands representing a novel approach for land cover classification using optical satellite data. Due to the adaptive nature of the algorithm, this approach may be quite useful for the development of a cost-effective land cover or agriculture monitoring system.