Improving the Quality and Spatial Resolution of Super-spectral and Hyper-spectral Images through Sensor Fusion

点击翻译按钮获取中文摘要

As much as 80 percent of the information used for decision making has some geospatial context. Multi-spectral satellites (with up to 10 spectral bands) have been increasingly launched in recent years for quickly capturing information of the earth's surface. It has been proven that super-spectral (10 to 50 spectral bands) and hyper-spectral (50 to 250 bands) remote sensing can provide much more information than multi-spectral remote sensing, because of their higher spectral resolutions (more spectral bands). *** However, due to the inherent limitation of electro-optic sensor physics, it is impossible to increase the number of the spectral bands of a given sensor without losing its image quality or spatial resolution or both. Therefore, very few super-spectral or hyper-spectral satellite missions have been launched to date. *** The proposed research will develop an alternative, sensor fusion solution to overcome the limitation of sensor physics. The solution will use or simulate two or more sensors to capture complementary information respectively (i.e. either high quality, high spatial resolution but low spectral resolution information, or high quality, high spectral resolution but low spatial resolution information), and then fuse the complementary information together to produce a super-spectral or hyper-spectral image that contains high quality, high spatial resolution and high spectral resolution information. *** New algorithms and software systems will be developed in the proposed research. The super-spectral images from WorldView-3 satellite and hyper-spectral images from EO-1 satellite will be used as experiment data. WorldView-3 was launched in August 2014. EO-1 satellite was launched in 2000 as part of NASA's New Millennium Program to enable the development of future earth imaging technologies. They are excellent data source for the proposed research. *** Based on our international success in increasing the spatial resolution of multi-spectral sensors through the fusion of the information from a panchromatic sensor (with 1 broad spectral band) and a multi-spectral sensor (with up to 8 spectral bands), it is anticipated that our new solution will be able to increase the spatial resolution of the super-spectral images of WorldView-3 (with 16 spectral bands), and increase the quality and spatial resolution of the hyper-spectral images of EO-1 (with 220 spectral bands). *** The successful outcome from this research will improve the image quality and spatial resolution of existing sensor systems. More importantly, it will contribute to the advancement of future Canadian and international super-spectral and hyper-spectral space missions. *****