Research And FPGA Verification Of Hyperspectral Image Endmember Extraction Algorithm

The development of endmember extraction technology plays an important role in the rapid analysis of hyperspectral image. At present, the more popular endmember extraction algorithms are mostly based on the PC platform, its advantages are easy to implement and disadvantages are the slow speed of extraction, poor real-time, thus it can not meet the requirements of rapid identification of ground objects. In order to overcome the above shortcomings, this design adopts FPGA to realize the automatic extraction of endmember。 The main task of the design are as follows:Firstly, the characteristics of hyperspectral data were analyzed. The advantages, and existing disadvantages or problems of extracting endmembers arithmetic were summarized. Then by contrast, fast PPI algorithm (FPPI algorithm) was proposed on the basis of the existing PPI algorithm. The algorithm principle and implementation process were introduced in detail. Finally, extraction algorithm FPPI in hardware on the FPGA was implemented. Through endmember extraction of a real hyperspectral remote sensing image, the algorithm FPPI achieved the intended purpose was proved.