UNIFIED HYPERSPECTRAL IMAGING METHODOLOGY FOR AGRICULTURAL SENSING USING SOFTWARE FRAMEWORK
Recently, the several studies employed hyperspectral imaging for agricultural remote sensing. A hyperspectral image contains enormous spectral waveband information in each pixel. Especially, ground-based hyperspectral imaging can obtain a high-resolution image with both spatial and spectral information. The objective of this study was to develop unified methodology for agricultural hyperspectral imaging. The methodology was based on the low-cost image capturing system and the object- oriented software framework developed in this study, and it can be employed for multipurpose use. First, in order to acquire a hyperspectral image, two image- capturing systems were developed. The portable image capturing system is a small system equipped with an electric driven pan head and can be used both in-room and in outdoor fields. The field-scale image capturing system with running vehicle can acquire large-scale images in-field. Then, a versatile software framework for hyper¬spectral imaging was designed and developed. The methodology in this study can be applied to various hyperspectral imaging projects. For each project, an individual estimation model has to be developed. Therefore, the methodology also included the concept of model development. Moreover, versatile script code library for model creation was developed with statistical computing software R. This study also showed the application examples of hyperspectral imaging projects that tried several agri¬cultural sensing.
Okamoto, H., Suzuki, Y., Kataoka, T. and Sakai, K. (2009). UNIFIED HYPERSPECTRAL IMAGING METHODOLOGY FOR AGRICULTURAL SENSING USING SOFTWARE FRAMEWORK. Acta Hortic. 824, 49-56
machine vision, image processing, camera, multivariate analysis, remote sensing