MORPHOLOGICAL ENDMEMBER IDENTIFICATION AND ITS SYSTOLIC ARRAY DESIGN

Abstract

Systolic arrays can be used in many different applications in order to improve performance. In particular, digital image processing algorithms are suitable to be implemented by systolic structures, since basic image manipulation operations are usually repetitive and can be mapped into a rectangular systolic structure. In this chapter we discuss the application of systolic arrays to speed up the performance of a new algorithm that performs unsupervised analysis of remote sensing images with high dimensionality (hyperspectral images). The methodology is based on mathematical morphology, which is a non-linear image analysis and pattern recognition technique that has proved to be especially well suited to segment images with irregular and complex shapes, but has rarely been applied to the classification/segmentation of hyperspectral images. The proposed method, which integrates spectral and spatial information in the analysis process, is fully described in this chapter, along with its hardware implementation by systolic arrays. A comparison between the hardware implementation and the software-based implementation is addressed.