 |
 |
 |
 |
||
 | |
 |
Projects (or Lab News) |
| ||||
 |
Analysis of Hyperspectral Images via Sparse Representations | |||
|
Date: | 11/07/2011 10:12 | ||
|
||||
|
In association with: | |||
|
Supervisor: | Asaf Elron | ||
|
Students: | TBD | ||
|
Background: In ordinary color imaging, a scene is represented using three spectral bands (red, green, blue), capturing the reflectance, or in some applications fluorescence, of the scene in three wide and overlapping spectral ranges. In hyperspectral imaging (HSI), the scene is represented by many (from tens to many hundreds) narrow, non-overlapping spectral bands, ranging from infrared, through the visible light to ultraviolet. This allows for different materials or substances in the scene to be distinguished from each other by using their unique reflectance or fluorescence spectra, making HSI a powerful tool. Hyperspectral imaging has many applications, ranging from earth sciences, particularly environmental studies, through biomedical research and diagnostics, to military applications. Each scene contains only a fixed verity of substances, each having a unique spectral signature. Thus, the spectra of the spatial pixels in the corresponding hyperspectral image are all combinations of these spectral signatures. Moreover, since the scene is generally piece-wise smooth, it is fair to expect that, for each spatial pixel, the spectrum be a combination of only a few of the available spectral signatures. This understanding implies much on the structure of hyperspectral images and thus assists in various analysis tasks involving them. When the spectral signatures observable in the scene are not known a prior, it is necessary to derive them from the image. Recently, improved methods for deriving the spectral signatures from the hyperspectral image have been demonstrated, spurring further research in the field. |
|||
|
Definition: In this project we will study two of the aforementioned methods for deriving the spectral signatures from a hyperspectral image. We will develop tools, methods and criterions for comparing the two methods and investigating their behavior. Finally, we will suggest, and possibly implement, an improvement to one of the methods or to both.
משך: חד-סמסטריאלי, ניתן להרחבה סביבת עבודה: MATLAB קדם: מבוא לעיבוד ספרתי של אותות מומלץ בצמוד: עיבוד ספרתי של אותות, הצגות דלילות ויתירות קדם לסטודנטים ממדעי המחשב: עיבוד אותות ותמונות במחשב
|
|||
|
Additional link: |
Learning Sparse Codes for Hyperspectral Imagery
| Return |  |