Satellite Image Based Dual Level Feature Approximation Technique for Efficient Sugarcane Monitoring Using ANN

Abstract

The problem of plant growth and yield estimation on sugarcane has been well studied. There exist numerous approaches towards the problem which consider only limited features like temperature, rainfall and so on. The methods suffer with poor accuracy in estimating growth and yield of sugarcane plants. To solve this issue, a dual level feature approximation model (DLFAM) has been presented in this article. The method uses both satellite and field images with other environmental features like temperature, humidity and hydrology features like water poured and rain fall.  Also, the features of ground like soil type, area of cultivation are considered. Using all these features, the method estimates support on low level features towards plant growth and yield according to color, contrast features. Also, the method computes the support on high level features according to different hydrology, environmental and other features. Using all these support values, the method computes the value of plant growth and yield towards efficient monitoring. Such features are trained with artificial neural network which classify the test features and returns the weight towards different class of sugarcane plants. The method introduces higher performance in plant growth and yield estimation.