Crop Yield Estimation Based on Unsupervised Linear Unmixing of Multidate Hyperspectral Imagery

Abstract

Hyperspectral imagery has the potential to describe the biological and chemical attributes of the plant in a better way compared to multispectral imagery because Hyperspectral imagery contains hundreds of spectral bands. This project describes Crop Yield Estimation Based on Unsupervised Linear Unmixing of Multidate Hyperspectral Imagery. In Hyperspectral imagery, the spectrum of each pixel is considered as the combination of the spectra of both vegetation and bare soil. By using Hyperspectral imagery technique the spectra of the vegetation and the bare soil images are extracted automatically. Based on the extracted spectra the vegetation abundances are computed. To reduce the influence of this uncertainty and to obtain robust estimation results the vegetation abundances that are extracted on two different dates on the same fields are combined. The experiment is carried out on the Multidate Hyperspectral images taken from two-grain sorghum fields. The result obtained from the correlation coefficient of vegetation abundances obtained by using unsupervised linear unmixing approaches are good compared to supervised learning where the vegetation and the bare soil are measured in the laboratory. In addition, the combination of vegetation abundances extracted on different dates can improve the correlations (from 0.6 to 0.7).

Existing System

In the existing system, artificial neural networks (ANNs) for the development of in -season yield mapping and forecasting systems was examined. The fertilization rates and various weed management protocols, were acquired by a compact airborne spectral imager. Statistical and ANN approaches along with various vegetation indices were used to develop yield prediction model. The high potential usefulness of ANNs was confirmed, particularly in the creation of yield maps.

Proposed System

In the proposed system, the hyperspectral imagery provides the better result than the multispectral imagery. Linear unmixing approaches are used for the crop yield estimation. Among these approaches, we use N-Finder and VCA methods which are the most efficient and the most widely used. N-Finder searches the simplex embedded inside the data sets, of which the volume is maximal. In the proposed system, the hyperspectral imagery provides the better result than the multispectral imagery. Linear unmixing approaches are used for the crop yield estimation. Among these approaches, we use N-Finder and VCA methods which are the most efficient and the most widely used. N-Finder searches the simplex embedded inside the data sets, of which the volume is maximal.

Architecture

ARCHITECTURE DIAGRAM