Publication:
Gpu-based implementation of target detection algorithms for hyperspectral images using nvidiar cuda
Gpu-based implementation of target detection algorithms for hyperspectral images using nvidiar cuda
dc.contributor.advisor | Vélez-Reyes, Miguel | |
dc.contributor.author | Trigueros-Espinosa, Blas | |
dc.contributor.college | College of Engineering | en_US |
dc.contributor.committee | Hunt, Shawn D. | |
dc.contributor.department | Department of Electrical and Computer Engineering | en_US |
dc.contributor.representative | Castellanos, Dorial | |
dc.date.accessioned | 2019-05-14T18:22:47Z | |
dc.date.available | 2019-05-14T18:22:47Z | |
dc.date.issued | 2011 | |
dc.description.abstract | Recent advances in hyperspectral imaging sensors allow the acquisition of images of a scene at hundreds of contiguous narrow spectral bands. Target detection algorithms try to exploit this high-resolution spectral information to detect target materials present in a scene, but this process may be computationally intensive due to the large data volumes generated by the hyperspectral sensors, typically hundreds of megabytes. Previous works have shown that hyperspectral data processing can significantly benefit from the parallel computing resources of GPUs, due to their highly parallel structure and the high computational capabilities that can be achieved at relative low costs. In this work, we studied the parallel implementation of target detection algorithms for hyperspectral images in order to identify the aspects in the structure of these algorithms that can exploit the parallel computing resources of GPUs based on the NVIDIA⃝R CUDATM architecture. A dataset was generated using a SOC-700 hyperspectral imager to evaluate the performance and detection accuracy of the parallel implementations. In addition, a library of target detectors was developed to facilitate the use of the algorithms by future researchers. | en_US |
dc.description.abstract | Avances recientes en los sensores hiperespectrales permiten la adquisición de imágenes de una escena a cientos de bandas espectrales contiguas y estrechas. Los algoritmos de detección tratan de aprovechar esta alta resolución espectral para detectar materiales de interés en una escena, pero este proceso puede ser computacionalmente intenso debido al gran volumen de datos generado por los sensores hiperespectrales, típicamente cientos de megabytes. Trabajos previos han mostrado que el procesamiento de datos hiperespectrales se puede beneficiar significativamente de los recursos de computación en paralelo de los GPUs, debido a su estructura altamente paralela y las altas capacidades de computación que pueden alcanzar a un precio relativamente bajo. En este trabajo estudiamos la implementación en paralelo de algoritmos de detección para imágenes hiperespectrales con el fin de identificar aspectos en la estructura de estos algoritmos que puedan sacar ventaja de los recursos de computación paralela de GPUs basados en la arquitectura CUDATM de NVIDIA⃝R . Un conjunto de datos fue generado usando una cámara hiperespectral SOC-700 para evaluar el rendimiento y la precisi ́on en detección de las implementaciones. En adición, se desarrollo una librería de algoritmos de detección para facilitar el uso de los algoritmos por futuros investigadores. | en_US |
dc.description.graduationSemester | Summer (3rd Semester) | en_US |
dc.description.graduationYear | 2011 | en_US |
dc.description.sponsorship | U.S. Department of Homeland Se- curity under award number 2008-ST-061-ED0001 and used facilities of the Bernard M. Gordon Center for Subsurface Sensing and Imaging Systems sponsored by the Engineering Research Centers Program of the National Science Foundation under Award EEC-9986821. Partial support was also received from the National Science Foundation under award EEC-0946463. | en_US |
dc.identifier.uri | https://hdl.handle.net/20.500.11801/2203 | |
dc.language.iso | English | en_US |
dc.rights.holder | (c) 2011 Blas Trigueros-Espinosa | en_US |
dc.rights.license | All rights reserved | en_US |
dc.title | Gpu-based implementation of target detection algorithms for hyperspectral images using nvidiar cuda | en_US |
dc.type | Thesis | en_US |
dspace.entity.type | Publication | |
thesis.degree.discipline | Computer Engineering | en_US |
thesis.degree.level | M.S. | en_US |
Files
Original bundle
1 - 1 of 1
- Name:
- ICOM_TriguerosEspinosaB_2011.pdf
- Size:
- 2.36 MB
- Format:
- Adobe Portable Document Format
- Description: