Professor Qian Du from Mississippi State University visited the Forum of IGG

(Correspondent Liu Huize)In the afternoon of December 16th, at the invitation of Dean Xiong Xiong and Vice Dean Wu Ke of the Institute of Geophysics & Geomatics, Professor Qian Du from the School of Electronic and Computer Engineering of Mississippi State University, USA, visited our college, and made a "Patch Tensor-based Hyperspectral Image Dimensionality Reduction and Classification" academic report in the 222 Academic Lecture Hall of the Wutan Building. More than 40 teachers and students attended the lecture.

Firstly, Professor Qian Du briefly introduced the advantages of hyperspectral images and some of the problems faced in processing hyperspectral images. Then, aiming at the problem of high spectral dimension, a solution was proposed, which was mainly to reduce the data volume by reducing the dimension. After that three dimensionality reduction techniques based on patch tensor were introduced, including improved tensor local preserving projection, low-order discriminant embedding of tensor and multi-map embedding based on patch tensor. Compared with the latest algorithms based on hyperspectral remote sensing images, the effectiveness of the method based on patch tensor was verified. Finally, it was concluded that combining spectral and spatial information to process hyperspectral images could effectively improve the accuracy of image classification.

During the discussion of the report, Professor Qian Du and the teachers and students present exchanged views on relevant frontier issues in hyperspectral remote sensing image processing, high score image processing, and agricultural remote sensing research.