Hyperspectral Camera Design Project Preliminary Design Report Project #P06522

Team Members: Will Shaffer Jeff Sidoni Dan Scorse Jordan Gartenhaus Sponsored by:

D3 Engineering 222 Andrews Street Rochester, NY 14604

D3 Engineering

Hyperspectral Camera Project (P06552)

Table of Contents 1. 2. 3. 4.

List of Figures………………………………………………………… 4 List of Equations ………………………………………………………5 Background of Sponsor ……………………………………………….6 Hyperspectral Introduction and Theory ………………………………7 4.1. Background of Hyperspectral Imaging …………………………..7 4.2. Spectroscopy and Spectral Reflectance …………………………..7 4.3. Hyperspectral Data Acquisition …………………………………..8 4.3.1. Linear Array Spectrograph ………………………………..9 4.3.2. Two Dimensional Sensor Method …………………………9 4.3.3. Data Cube Composition …………………………………10 4.4. Identification and Classification of Materials …………………..11 4.5. Applications of Hyperspectral Imaging …………………………12 5. Project Outline ……………………………………………………….13 5.1. Purpose ………………………………………………………….13 5.2. Requirements ……………………………………………………13 5.3. Project Timeline …………………………………………………14 6. Preliminary Design …………………………………………………..15 6.1. Imaging System …………………………………………………15 6.1.1. CCD vs. CMOS Imagers …………………………………15 6.1.2. Texas Instruments C6416 vs. DM642 DSP ………………16 6.1.3. Preliminary Imaging System Design Components ………17 6.2. User Interface ……………………………………………………17 6.3. Optical System Design …………………………………………17 6.3.1. Background Motivation ………………………………….17 6.3.2. Abstract ………………………………………………….17 6.3.3. Preliminary Research …………………………………….18 6.3.4. ImSpector Research ………………………………………20 6.3.5. Alternate Dispersion Element Design ……………………21 6.3.6. Prism Design ……………………………………………..24 6.3.7. Final Optical Design ……………………………………..29 6.4. Scanning Mirror Control ………………………………………..30 6.4.1. System Overview ………………………………………30 6.4.2. Stepper Motor Control …………………………………..31 6.4.3. Design Requirements …………………………………….32 6.4.4. Slit Width Analysis ………………………………………34 6.4.5. Kruse Control …………………………………………….36 6.5. Data Normalization and Material Identification ………………..38 6.6. Mechanical System and Enclosure ………………………………39 6.6.1. Requirements …………………………………………….39 6.6.2. Mounting Hardware ……………………………………39 6.6.2.1. Scanning Mirror Mount …………………………….39 6.6.2.2. ImSpector Lens Mounting System …………………40 6.6.2.3. Custom Design Lens Mounts ……………………….40 6.6.2.4. Imaging Device Mount …………………………….40

PDR

definition • design • development

Page 2 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.6.3. System Enclosure ………………………………………40 6.7. Preliminary Bill of Materials ……………………………………42 7. Design Feasibility and Expected Technical Issues …………………..43 8. References ……………………………………………………………44

PDR

definition • design • development

Page 3 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

1. List of Figures:

PDR

Cover Image: Figure 4.1: Figure 4.2: Figure 4.3: Figure 4.4: Figure 4.5: Figure 4.6: Figure 4.7: Figure 4.8:

Hyperspectral Data Cube Electromagnetic Spectrum ………………………….7 Spectral Signature of Various Materials……………..8 Linear array Imaging Spectrometer ………………….9 Two-dimensional Imaging Spectrometer …………..10 Hyperspectral Image Cube …………………………11 Panchromatic Combat Area Image …………………12 Multispectral Combat Area Image………………….12 Hyperspectral Combat Area Image…………………12

Figure 5.1:

Project Timeline ……………………………………14

Figure 6.1: Figure 6.2: Figure 6.3: Figure 6.4: Figure 6.5: Figure 6.6: Figure 6.7: Figure 6.8: Figure 6.9: Figure 6.10: Figure 6.11: Figure 6.12: Figure 6.13: Figure 6.14: Figure 6.15: Figure 6.16: Figure 6.17: Figure 6.18: Figure 6.19: Figure 6.20: Figure 6.21: Figure 6.22: Figure 6.23: Figure 6.24: Figure 6.25: Figure 6.26: Figure 6.27:

System Block Diagram ……………………………..15 Imager Response Curve …………………………….16 90 Degree Optical Path ……………………………..19 Straight through Optical Path……………………….20 Prism Grating Prism Assembly …………………….21 Littrow Prism Spectrograph ………………………..22 Littrow Prism Spectrograph with mirror……………23 Wadsworth constant-deviation mounting …………..23 Littrow-mounted Prism Spectrograph ……………..24 Geometry of a Prism ………………………………..24 Geometry of Focal Length and Detector Height …..27 Lambda vs. Height …………………………………27 Wavelength vs. Optimized Refracted Angle ……….27 Detector Size vs. Input Angle ………………………28 Detector Size vs. Prism Angle………………………28 Pugh Diagram of Optical Design …………………..29 Phase ‘A’ Driver Schematic ………………………..31 Two Phase Stepper Motor Configuration ………….31 Micro-step Phase Winding Signals ………………..32 Open-Loop Stepper Motor Control…………………32 Required Target Workspace ………………………..33 Scanning Mirror Parameters ………………………..34 Image Slit Width Analysis………………………….35 Two Phase Motor Schematic and Model …………..37 Closed-loop Kruse Stepper Motor Controller………37 Slit Width Analysis for Kruse Controlled System …38 Preliminary Bill of Materials ……………………….42

definition • design • development

Page 4 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

2. List of Equations Equation 6.1: Equation 6.2: Equation 6.3: Equation 6.4: Equation 6.5: Equation 6.6: Equation 6.7: Equation 6.8-11: Equation 6.12: Equation 6.13-23:

PDR

Optical Path Length ………………………….18 Index of Refraction of BK7…………………..25 Exit Angle ……………………………………25 Detector Size …………………………………25 Change in Exit Angle ………………………26 Target Image Size ……………………………34 Slits per Scan…………………………………34 Successive Slit Width Calculations…………..34 General Slit Width Formula ………………….35 Kruse Control Functionality Overview …..36-37

definition • design • development

Page 5 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

3. Background of Sponsor D3 Engineering provides DSP Hardware, Software, and Algorithms in signal processing applications. The range of projects includes medical, wireless, image processing, and motor control.

The Hyperspectral Camera is the 3rd RIT Senior Design project that D3 Engineering has sponsored. The Senior Design Project is used by D3 to evaluate potential full-time employees, develop IP, and prepare prototypes for potential customer demonstration.

PDR

definition • design • development

Page 6 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

4. Hyperspectral Introduction and Theory 4.1 Background of Hyperspectral Imaging Imaging in its traditional sense has always been though of as taking pictures using three distinct wavelengths of light; red, green and blue. In remote sensing, the term hyperspectral refers to an imaging system capable of capturing up to several hundred narrow, contiguous spectral bands of the electromagnetic spectrum, as seen in Figure 4.1.

This method of acquiring data is what

distinguishes a hyperspectral system from a traditional multispectral system, where only a pre-selected set of wavelengths are captured. This ability allows for the detection of subtle variations that are often overlooked by less informative multispectral and traditional imaging methods.

The interest in this type of

imaging has increased greatly due to the amount of information that an image contains.

Figure 4.1 – The Electromagnetic Spectrum

Advancements in manufacturing allow for better imaging systems, imagers with lower signal to noise ratios and more accurate methods of calibrating the cameras.

4.2 Spectroscopy and Spectral Reflectance Spectroscopy is, by definition, the study of the absorption or reflection of various wavelengths of light by a substance. There are three main types of

PDR

definition • design • development

Page 7 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

spectroscopy; emission, scattering and absorption. This type of imaging uses the latter to measure the amount of light reflected from the materials under investigation. The spectral reflectance of an object is the ratio of reflected energy to incident energy as a function of wavelength. The reflectance spectrum, also known as a hyperspectral signature, of a material is a plot of this spectral reflectance with respect to wavelength. Figure 4.2 gives an example of a spectral signature for several materials.

70 Asphalt Dry Grass Granite Grass Dark Soil

60

% Reflectance

50 40 30 20 10 0 -10

0

5

10

15

Wavelength λ (μm)

Figure 4.2 – Example of Spectral Signature

4.3 Hyperspectral Data Acquisition Hyperspectral images are obtained using a device called an imaging spectrometer. By definition, this is an optical device used to measure various properties of light over the electromagnetic spectrum. Typically, it contains an optical system, a dispersing element such as a prism or grating, and an array of detectors that are designed to operate over a wide variety of wavelengths.

PDR

definition • design • development

Page 8 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

4.3.1 Linear Array Spectrograph A linear array spectrograph contains a single row of detectors, depicted by the simplified diagram in Figure 4.3. This figure shows how the spectral plot of a single cell of the target is obtained. The reflected light from this cell is broken into is spectral content by the element, and the array of detectors measures the respective values of the intensity of the incident light.

Figure 4.3 – Simplified Imaging Spectrometer

In order to obtain a complete row of the target, it is necessary to use a mirror that scan across the target while the spectrograph evaluates the spectral content of each individual cell in the row. The limitation of this setup is the contingency that another element is responsible for the second dimension of motion over the target. In remote sensing applications this is usually accomplished by the movement of the camera inside a flying plane or a non-geosynchronous satellite over the target area. 4.3.2 Two Dimensional Sensor Method An alternative method of capturing the data is to replace the linear array with a two dimensional CMOS or CCD imager. Figure 4.4 shows a simplified diagram of such a system. One row of the target is passed through a slit in the front end optical system and into the dispersion element. The output of the

PDR

definition • design • development

Page 9 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

element is a two dimensional representation of the spectrum of each cell of the row, which the sensor detects. One axis of the sensor represents the spectral content, while the other represents the spatial.

Figure 4.4 – Two Dimensional Sensor Method

The use of a two dimensional sensor allows the imaging system to cover a single linear dimension with one frame of the imager, while the linear array spectrometer has to take successive frame captures to create a row. This will simply the data acquisition in a lab based environment has the possibility of eliminating the need for moving mechanical components, such as the scanning mirror, in remote sensing applications where a two dimensional target is acquired. 4.3.3 Data Cube Composition The final data set consists of successive images with high spectral resolution that allows for materials to be identified, unlike old systems that could only differentiate between materials. In order to organize the data in a way that is understandable, the successive two dimensional images acquired are combined to form a three dimensional data set, known as a hyperspectral cube. Figure 4.5 shows how a data cube is comprised; two spatial axes and one spectral axis. PDR

definition • design • development

Page 10 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Figure 4.5 – Hyperspectral Image Cube

4.4 Identification and Classification of Materials Due to the vast amounts of information contained in a hyperspectral data set, the spectral responses of the pixels can be used to identify the materials present. There are several libraries available that contain plots of spectral signatures for hundreds or even thousands of materials ranging from natural to man-made. Several issues arise when extracting the data from a raw image cube. In remote sensing applications, the target cell that is being observed rarely contains reflectance values due to a single material since the target cell size is large. While the degree of distortion is dependent on exactly how large the cell size is, the resulting spectra is a combination of all the different materials within that cell, resulting in a composite or mixed spectrum. Classification and identification pose other problems when compared directly with a reference spectral signature plot; the values measured with the imager correspond to the radiance values off the surface of the material, and must be converted into relative reflectance before a direct comparison can be made. Several methods to accomplish this are available; some of which require only the data set, and others require some knowledge about the conditions in which the images were taken. Using the approach of trying to match the acquired data set directly with a library requires an accurate conversion from radiance to reflectance. Since the capture spectra will not match exactly, it must be compared and the correlation between the reference and the observed, rated, and from this rating a decision

PDR

definition • design • development

Page 11 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

about the identity of the material can be made. This is complicated by issues discussed previously with mixing of different spectra, a topic that will be investigated further as the project progresses. 4.5 Hyperspectral Applications With the amount of information present in a hyperspectral data cube, there are an abundance of applications ranging from medical to agricultural to defense. Remote sensing hyperspectral cameras are currently being used by the military to detect targets and are normally hidden to traditional color and even multispectral imaging systems. Figures 4.6-4.8 below show the same image viewed with these three types of systems. The target being observed is a combat zone which contains a camouflaged tank.

Figure 4.6-4.8 – Defense Remote Sensing Application

Other remote sensing applications include forest heath observations for fire prevention, tracking vegetation, measuring soil quality and terrain mapping. Several applications of hyperspectral imaging that are on a smaller scale occur in the medical industry for early cancer detection and food borne illness detection and prevention.

Additionally, this technology is being

utilized by the government to identify counterfeit currency.

PDR

definition • design • development

Page 12 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

5. Project Outline 5.1. Purpose The purpose of this design is to build a functional hyperspectral camera system for use by D3 Engineering’s design staff. 5.2. Requirements The design requirements for this project were left very open, which allows for a very research oriented design project. The following list defines the requirements specified by the customer. •

The hyperspectral camera shall use a single CMOS or CCD detector.

•

The hyperspectral camera shall have a spectral resolution of 25 to 50nm over the visible region; 400-850nm.

•

The hyperspectral camera shall utilize a scanning mirror to acquire images.

•

The hyperspectral camera shall use pre-existing hardware developed by D3 Engineering for the image capture and motion control.

PDR

definition • design • development

Page 13 of 44

PDR

Design Documentation

Initial Prototyping w / CDK

Working Camera

Preliminary Design Review

CDK

3

4

5

6

7

Frequency registration

Scanning Mirror

17

definition • design • development

Final Report

Verify vs Requirements

31

32

Testing

30

Optical Mounts

28

Intergration

Enclosure

27

29

Mirror

Create data Cube

24

Mechanical

Frequency Resolution

23

26

Conversion to Matlab

25

Raw Data Dow nlaod

22

Softw are

21

20

Mechanical Structure

Testing

16

19

20 days

Simulation

15

Linear stepper motor

2 w ks

Lightpath Diagram

14

1 wk

1 wk

3 w ks

20 days

4 w ks

4 w ks

2 w ks

50 days

3 w ks

3 w ks

2 w ks

1 wk

45 days

4 w ks

4 w ks

2 w ks

2 w ks

2 w ks

Optical Design

13

7 w ks

40 days

2 w ks

1 wk

9.2 w ks

61 days

1 day

2 w ks

Fri 5/19/06

Mon 5/15/06

Mon 4/24/06

Mon 4/24/06

Mon 4/10/06

Mon 3/13/06

Mon 2/27/06

Mon 2/27/06

Mon 4/10/06

Mon 3/20/06

Mon 3/6/06

Mon 2/27/06

Mon 2/27/06

Mon 2/27/06

Mon 2/27/06

Mon 2/27/06

Mon 4/10/06

Mon 3/27/06

Mon 3/13/06

Mon 2/27/06

Mon 2/27/06

Mon 2/27/06

Tue 3/14/06

Tue 3/7/06

Mon 1/2/06

Mon 1/2/06

Fri 2/24/06

Mon 1/2/06

2 w ks Mon 12/19/05

Use Spectrograph Optics

18

Mon 12/5/05

Mon 12/5/05

Start

2 w ks Mon 12/19/05

12

Optical Subsystem

10

11

1Mpixel

9

Camera Softw are Development

Requirements Generation

2

8

124 days

Hyperspectral Camera Design

1 2 w ks

Duration

Task Name

ID

Thu 5/25/06

Fri 5/19/06

Fri 5/12/06

Fri 5/19/06

Fri 5/5/06

Fri 4/7/06

Fri 3/10/06

Fri 5/5/06

Fri 4/28/06

Fri 4/7/06

Fri 3/17/06

Fri 3/3/06

Fri 4/28/06

Fri 3/24/06

Fri 3/24/06

Fri 3/24/06

Fri 4/21/06

Fri 4/7/06

Fri 3/24/06

Fri 3/10/06

Fri 4/14/06

Fri 4/21/06

Mon 3/27/06

Mon 3/13/06

Mon 3/6/06

Mon 3/27/06

Fri 2/24/06

Fri 1/13/06

Fri 12/30/05

Fri 12/30/05

Fri 12/16/05

1/2

Dec '05 Jan '06 Feb '06 Mar '06 Apr '06 May '06 Jun '06 4 11 18 25 1 8 15 22 29 5 12 19 26 5 12 19 26 2 9 16 23 30 7 14 21 28 4 11 Thu 5/25/06

Finish

D3 Engineering Hyperspectral Camera Project (P06552)

5.3. Project Timeline Figure 5.1 shows the projected timeline for this project.

Page 14 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6. Preliminary Design Since this project is more research oriented than developing a final product, there are minimal design constraints. The block diagram in Figure 6.1 depicts a high-level model of the design.

irr o M

Imager0

Optics

r

Scan Control

Figure 6.1 – System Block Diagram

6.1. Imaging System The imaging system will consist of an embedded platform that will interface to an image sensor capable of capturing data over the specified bandwidth of 400 to 850nm. 6.1.1. CCD vs. CMOS Imagers CCD (charge coupled devices) and CMOS (complementary metal oxide semiconductor) imagers both provide a way to digitally capture images, but each have their respective benefits. CMOS imagers provide a simplified method of interfacing due to the integration of all the timing circuitry necessary to clock data out of the sensor, while CCD sensors require a significant amount of external circuitry to obtain the data. Key advantages to a CMOS sensor include a lower power usage, this integration of timing circuitry, and lower cost. The most significant property in this application is image quality, where the CCD surpasses that of a CMOS sensor. While this will become an issue in the

PDR

definition • design • development

Page 15 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

future, the main goal of this project is to develop a working hyperspectral camera prototype, and since D3 Engineering already the ability to interface to a CMOS sensor, this technology will be utilized in the beginning stages of this design. The bandwidth specification of this project is between 400 and 850nm, which as seen by the spectral response curve of the chosen imager in Figure 6.2, will be adequate for this specification. Future improvements of this project may include the use of a CCD image sensor over a CMOS and the possibly of migrating to a 12 or 14 bit sensor.

Figure 6.2 – Spectral Response of Imager

6.1.2. Texas Instruments C6416 vs. DM642 DSP There are two digital signal processors available from Texas Instruments that were considered during this design, namely the C6416 and the DM642. D3 Engineering has hardware to interface to both processors, but connectivity to a PC varies by platform.

The system will be connected using a USB interface;

therefore due to hardware limitations the C6416 will be used. While the DM642 is TI’s flagship digital media processor, the complication of the video port driver used to capture images may cause unnecessary problems syncing the movement of the scanning mirror with the frame capture. The C6416 has a higher clock frequency, running at 1GHz, than the DM642 which processes at 720MHz. At the early stage of development the processing speed is of little concern since the algorithm development will be done in MATLAB. PDR

definition • design • development

Page 16 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.1.3. Preliminary Imaging System Design Components The design will utilize hardware that D3 Engineering will provide to capture images using an embedded system. The final imaging system specifications are as follows: •

1.3 Mega Pixel 10-Bit CMOS Image sensor

•

D3 Engineering’s Camera Developers Kit for image acquisition.

•

Spectrum Digital 6416 DSK using a Texas Instruments C6416 1GHz Digital Signal Processor

6.2. User Interface In the early development phase of this project the post-processing will be preformed in MATLAB after the image cubes have been captured. This will ease the experimentation and development of the algorithms necessary to obtain spectral signatures and classify materials from the attained data sets.

The system will

interface with MATLAB over a high speed USB 2.0 connection. 6.3. Optical System 6.3.1. Background Motivation The desire to take on the optical design for this camera stemmed from the fact that the work experience of a team member, who has spent a year and half working for a company that, in the particular division, dealt mostly with telescope design, integration, and testing. The basic knowledge of optics and optical testing beyond how it applies to a telescope would be beneficial, especially for possible career paths. 6.3.2. Abstract The system had a wavelength requirement of 400-850 nm and a resolution of 25-50 nm. Size of the system is to be bench top, but portable. A scanning mirror will be integrated into the front of the system in order to sweep the target while the detector or data acquisition unit captures the images. Through thorough internet and library research, optical path designs and dispersion elements were considered. Such elements such as diffraction grating, wedge filters, and prisms were investigated while straight through and diffracted

PDR

definition • design • development

Page 17 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

optical paths were considered. Existing off the shelf designs from Edmund Optics and Specim were also researched and found to be feasible options for the given specifications. Low-end calculations using a prism as a dispersing element, along with correspondence with a representative of Specim, a company who produces a spectral photometer, discussions with Dr. Wells of ITT, and discussion’s with Edmund Optics technicians aided concept selection. The results of the above effort yielded an off the shelf component from either Edmund Optics or Specim as the best choice in order to provide the most effective prototype for the customer. 6.3.3. Preliminary Research Dr. Conrad Wells of ITT was approached with the problem statement for this design and an ensuing discussion took place. Dispersion elements such as grating and prisms were first brought to attention through this discussion as there had been no previous knowledge of how the wavelength range for a given resolution was to be achieved. Very low level, but necessary for understanding, light paths were illustrated to give a direction for what the system design may resemble. Minimal requirements for an optics system consist of three lenses, a variable slit and a dispersion element inside an enclosure. Figures 6.3 and 6.4 illustrate beginning comprehension of an optical system.

The figures also

illustrate other components of the optical system such as a variable slit, a scanning mirror, and baffles for stray light. Further research on the internet revealed a calculation for determining optical path length. A = ∫ n( s ) ds

(6.1)

c

Where ‘A’ represents path length and n(s) is index of refraction as a function of distance. The index is also dependent upon the glass element light is passing through.

PDR

definition • design • development

Page 18 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

MIRROR TARGET

LENS (3) PLACES BAFFLE

ENCLOSURE

PRISM

Figure 6.3 – 90 Degree Optical Path

PDR

definition • design • development

Page 19 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

MIRROR TARGET

LENS (3) PLACES BAFFLE

ENCLOSURE

PRISM

Figure 6.4 – Straight through Optical Path

6.3.4. ImSpector Research The complex nature of an optical system promoted research of off the shelf optical systems. Specim makes a number of ImSpector cameras that capture images and display data cubes as specified by the customer.

Initially, the

impression was that the ImSpector was only capturing visible or infrared light separately; however, that is not the case as search results yielded that the AISA Eagle is a bench top unit with dimensions 165mm x 200mm x 390mm and weighing no more than 6kg. This suggested that the design could realistically be

PDR

definition • design • development

Page 20 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

sized to bench top. Correspondence with Specim resulted in contact with Wes Procino, a sales representative who graduated from the University of Rochester. Procino was very helpful in answering questions and offering suggestions. Reading about the ImSpector camera revealed that a “straight through” optical design for all their cameras was discovered by utilizing a specific dispersion element, Prism-Grating-Prism. The Prism-Grating-Prism or PGP was developed by Mauri Aikio’s, whose dissertation was the creation of a Hyperspectral PGP Imaging Spectrograph. Aikio invented the PGP element as a low-cost option to make spectrographs and as a result of his work, a company in 1995 was founded to manufacture the PGP technology. The PGP element is a mate of a prism, long pass filter (LPF), covering glass, grating, substrate glass, short pass filter (SPF), and a second prism. Figure 6.5 depicts this. Covering glass

Prism 1

SPF

Prism 2

LPF

Grating

Substrate glass

Figure 6.5 – Prism-grating-prism element

6.3.5. Alternate Dispersion Element Design While the possible intellectual property held by the PGP system was being researched, alternate dispersion element designs were investigated. Simple prism designs were looked at and an array of calculations were done to assess the

PDR

definition • design • development

Page 21 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

feasibility of a custom optics design in this project. Prism designs along with alternate dispersion methods are assessed in subsequent sections. After calculations had been preformed on the prism design, more technical advise was obtained about the possible design. Holographic gratings were discussed, but the use of a monochromator is preferred. If the design utilized a holographic grating, a slit would be required and calibration procedures preformed.

The attraction of the monochromator’s holographic dispersion

element revolves around the pre-calibration of the unit off the shelf; a subject that is very complex. Design’s in Figures 6.6 – 6.9 all utilize a Littrow prism which was not discovered until late in the preliminary design process. A Littrow prism refracts the light into it and then reflects it back out into the path that it came in, effectively cutting the focal length into a fraction of a normal refracting prism. A silver or aluminum backing on the prism allows for this reduction in focal length.

Figure 6.6 – Littrow Prism Spectrometer

PDR

definition • design • development

Page 22 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Figure 6.7 – Littrow Prism Spectrometer with Mirror

Figure 6.8 – Wadsworth constant-deviation Mounting Device

PDR

definition • design • development

Page 23 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Figure 6.9 – Littrow-mounted prism spectrograph

6.3.6. Prism Design At first glance, a simple prism design seemed like it would be the cheapest and easiest dispersion element in an optical system. In order to prove this, calculations optimizing the focal length, detector size, prism angle, and angle the prism is held (or incident angle) had to be preformed. While the calculations were lengthy, it was necessary to properly optimize the parameters. Calculations were preformed for a range of wavelengths from 400 nm to 700 nm. Normal to surface

Normal to surface

α

ϑi1

δ

ϑt 2

Figure 6.10 – Geometry of a Prism

PDR

definition • design • development

Page 24 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

The prism material being used was BK7. The index of refraction for BK7 as a function of wavelength is as follows:

n = 1+

B1λ B 2λ B3λ + + λ − C1 λ − C 2 λ − C 3

(6.2)

Where B1, B2, B3, C1, C2, C3 are constants of the index of refraction. The angle of incidence was predetermined and changed as needed for optimization. Exit angle, ϑt 2 , out of BK7 in a function of the index of refraction, incidence angle, ϑi1 , and the prism angle, α . predetermined, but changed for optimization.

The prism angle was also

Usually, an angle between 45

degrees or 60 degrees is utilized.

ϑt 2 = a sin[(sin α )(n 2 − sin 2 ϑi1 )1 / 2 − sin ϑi1 cos α ]

(6.3)

Detector size is determined by simple trigonometry, as seen in Figure 6.11 and Equation 6.4 tan ϑt 2 =

h f

(6.4)

h

ϑt 2 f Figure 6.11 – Geometry of focal length and detector height

PDR

definition • design • development

Page 25 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

The variable ‘f’ represents the focal length and ‘h’ represents the height of the detector. Units are in millimeters. This is a function of the exit angle and thus a function of wavelength; for every wavelength there is a change in height, and the overall delta in height returns the detector size. Finally, a delta is found between the angle exiting the prism, ϑt 2 , and the normal to the prism in order to verify the calculations were done correctly.

δ = ϑi1 + sin −1 [(sin α )(n 2 − sin 2 ϑi1 )1 / 2 − sin ϑi1 cos α ] − α

(6.5)

The delta as a function of wavelength, seen in Equation 6.5, varied between 32 and 35 degrees. To add a degree of verification to the design, calculations were done with an incident angle of zero degrees into a 60 degree prism. The exit angle was 60 degrees with a delta of 30 degrees which was expected. The figures below depict results to these calculations as well as a copy of the spreadsheet used.

PDR

definition • design • development

Page 26 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

lamda vs. h 30

20

h (mm)

10

0 1 1. 05

0. 4 0. 45 0. 5 0. 55 0. 6 0. 65 0. 7 0. 75 0. 8 0. 85 0. 9 0. 95

Series1

-10

-20

-30 lamda (nm)

Figure 6.12 – Lambda vs. Height

Wavelength vs. Optimized Refracted Angle 0.03

0.01 0 1.05

1

0.95

0.9

0.85

0.8

0.75

0.7

0.65

0.6

0.55

0.5

0.45

Series1 0.4

Theta_t2-theta_not (deg)

0.02

-0.01 -0.02

-0.03 Lamda (nm)

Figure 6.13 – Wavelength vs. Optimized Refracted Angle

PDR

definition • design • development

Page 27 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Detector size Vs Input angle 100 90

Focal Plane Size (mm)

80 70 60 50

Series1

40 30 20 10 0 0

10

20

30

40

50

60

70

Incident Angle (deg)

Figure 6.14 – Detector Size vs. Input Angle

detector size vs. prism angle Series1

50

Series2 45

Series3

40

Series4

detector size (mm)

Series5 35

Series6 Series7

30

Series8

25

Series9

20

Series10 Series11

15

Series12

10

Series13 Series14

5

Series15 0 0

10

20

30

40

50

60

70

80

prism angle (deg)

Figure 6.15 – Detector Size vs. Prism Angle

PDR

definition • design • development

Page 28 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.3.7. Final Optical Design The following factors were considered for the final design decision: Results

from extensive calculations performed on utilizing a prism, alternate dispersion elements and optical path’s research, a Pugh diagram of design trades, seen in Figure 6.16, and the customer’s specifications. Specifically, the desire of the customer to investigate the data acquisition and image processing rather than the construction of an optical system. Design

Datum

Handheld Spectroscope

Prism

Grating & Slit

Chucks Apparatus

Impspector

Monochromator

Cost

$500

+ ($272)

+ ($200)

($900)

+

-

($1165)

Size

“Bench

+

S

S

S

S

+

Criteria

Top” Complexity

Ready off the shelf

+

assembly and calibration required

assembly and calibration required

design of light source input needed

+

+

Light source

White light/Sun light

S

S

S

-

S

S

Achieved Wavelength

400700nm

S

S

S

S

S

S

Achieved Resolution

25-50nm

+

+

+

TBA

+

+

Tuning

Desired wavelength can be chosen

-

-

-

+

Totals +=1 - = -1 S=0

+ +

3

1

0

0

2

3

Figure 6.16 – Pugh Diagram

Calculations performed for utilizing a prism where done with the focus of optimizing the focal length, incident angle, prism angle, and angle the prism is set at. All of these parameters were optimized at reasonable values with the exception of the focal length.

PDR

It was decided that 75-200 millimeters is an

definition • design • development

Page 29 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

optimal focal length and the length that came out of the analysis was approximately one meter. This was unacceptable for a bench top design of this system. Alternate forms of dispersion elements had to be looked into. After further research about alternate dispersion elements, a blazed holographic grating from Edmund Optics was discovered.

This dispersion

element could be set at an optimal focal length of 150mm. A discussion with Edmund Optics, however, suggested the use of a monochromator or a handheld spectroscope over a custom built optics system. Both items are already calibrated and are off the shelf ready to be implemented into a system. The final design recommendation is to use an off the shelf system from either Edmund Optics of Specim. In addition, a custom optics system will be experimented with to extend the customers knowledge about the subject of hyperspectral imaging. 6.4. Scanning Mirror Control 6.4.1. System Overview This project requires the use of a scanning mirror to focus light from

successive slits of a stationary target image into the optical system and eventually the CMOS imager.

Since the slit widths from the target image are ideally

dimensionless, it is advantageous to implement a motion control device with the highest possible resolution. As a starting point, a 100 pole-pair 2-phase stepper motor has been chosen to rotate the mirror which provides a 3.6° per full-step resolution. Illustrated in figure 6.17 is the schematic for the circuitry needed to drive one phase (A) of a stepper motor. An identical circuit is used to drive the second phase (B). Based upon this configuration, four PWM signals are required to drive the stepper motor and rotate the scanning mirror.

PDR

definition • design • development

Page 30 of 44

D3 Engineering

Hyperspectral Camera Project (P06552) Vmm 100 uF

ZMM5251B

IRFZ44E

IRFZ44E

100

100

Half - Bridge Driver

Half - Bridge Driver 10k

2 3

PWM_A1-

4 8 1 6

Vdd

5 7

IN LO SHDN HO

5 7

MtrDA1+

MtrDA1-

0

COM VB VCC VS

10k

0

LO IN HO SHDN

0 IRFZ44E

COM VB VCC VS

IRFZ44E

2 3

PWM_A1+

4 8 1 6

Vdd

0

0 1uF

IR2104

100

100 10k

820nF

0

0 .033

2

+

1

.033

0

47k -

ViA1

1uF

10k

1k

Motor Current Sense

IR2104

3

1k

820nF

47k

V3_3/2

Figure 6.17 – Phase ‘A’ Driver Schematic

6.4.2. Stepper Motor Control

Figure 6.18 – Two Phase Stepper Motor Configuration

There exist a number of methods to control the magnitude of the angle rotated per step. Full-stepping is a method by which both phase windings are energized with the polarity being switched by an alternating current and thus requiring four cycles to rotate by a full step. Using this method, the stepper motor chosen for this project would rotate 3.6° for each successive slit. Half-stepping doubles the resolution by alternating between single and dual-phase operation.

PDR

definition • design • development

Page 31 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Micro-stepping is a technique in which sinusoidal signals are used to drive the phase windings as opposed to discrete transitions. This method allows the motor’s natural step size to be further subdivided into anywhere from 16 to 256 micro-steps. Figure 6.19 illustrates the applied signals to the phase windings in a micro-step mode.

Figure 6.19 – Micro-step Phase Winding Signals

Each of the methods described above operate in an open-loop controller configuration. In section 6.4.5, a technique for calculating and feeding the rotor position back into the controller is outlined. A block diagram for the open-loop configuration is shown in Figure 6.20.

Figure 6.20 – Open-Loop Stepper Motor Controller

6.4.3. Design Requirements Implementing the stepper motor controller is going to be an incremental

process. The required number of slits per target image to generate a useful data set remains a variable in this design; and is largely a function of how accurately the stepper motor can be positioned. The minimum requirement for the scanning mirror is a +/- 20° range of motion with respect to the center of the target image. Figure 6.21 illustrates the target workspace.

PDR

definition • design • development

Page 32 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

Figure 6.21 – Required Target Workspace

As a baseline, an open-loop controller configuration will be implemented with 16 micro-steps per full step. This design generates an effective step angle of .225° and 177 total slits. The corresponding slit widths remain only a function of the distance between the scanning mirror and the target image. As mentioned above, the number of micro-steps will gradually increase as the controller design progresses until satisfactory image resolution is achieved.

Notice also from Figure 6.21 that as the scanning mirror rotates towards the edges of the target, the corresponding slit widths increase and must be accounted for in the image processing algorithm. Section 6.4.4 provides a more detailed analysis of this effect.

PDR

definition • design • development

Page 33 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.4.4. Slit Width Analysis

Figure 6.22 – Scanning Mirror Parameters

The size of the target image as a function of the distance between the target and the scanning mirror is given by:

Target Width = 2 y

y = d ⋅ tan(20°)

(6.6)

The number of slits per full 40° swing is given by:

nSlits 2 ⋅ 20° = Full Image ΔΘ

(6.7)

In order to develop a mathematical relationship between successive slit widths, start at the center of the image and rotate outwardly.

Δy1 = d ⋅ tan(ΔΘ)

(6.8)

Δy2 = d ⋅ tan(2ΔΘ) − Δy1

(6.9)

Δy3 = d ⋅ tan(3ΔΘ ) − Δy2 − Δy1

(6.10)

i −1

Δyi = d ⋅ tan(iΔΘ) − ∑ ( Δy j ) ∀ i > 1

(6.11)

j =1

The equation for the ith slit width can be simplified further with the following realizations:

PDR

definition • design • development

Page 34 of 44

D3 Engineering

Hyperspectral Camera Project (P06552) i −1

Δyi = d ⋅ tan(iΔΘ ) − ∑ ( Δy j ) j =1

i−2 ⎛ ⎞ i−2 = d ⋅ tan(i ΔΘ ) − ⎜ d tan ( (i − 1) ΔΘ ) − ∑ Δy j ⎟ − ∑ Δy j j =1 ⎝ ⎠ j =1 = d ⋅ tan ( iΔΘ ) − d ⋅ tan ( (i − 1) ΔΘ )

(6.12)

Based upon the equations derived above the following table is compiled to illustrate the relationship between the number of micro-steps per full step, step angle size, target size, and image slit width.

d [m]

y (half target width) [m]

1

0.36397023

2

0.72794047

3

1.09191070

4

1.45588094

5

1.81985117

μStep Size 0.0625 0.03125 0.015625 0.0078125 0.00390625 0.0625 0.03125 0.015625 0.0078125 0.00390625 0.0625 0.03125 0.015625 0.0078125 0.00390625 0.0625 0.03125 0.015625 0.0078125 0.00390625 0.0625 0.03125 0.015625 0.0078125 0.00390625

3.6° per Full Step: # Slits/ Δθ Half Image 0.225 88 0.1125 177 0.05625 355 0.028125 711 0.0140625 1422 0.225 88 0.1125 177 0.05625 355 0.028125 711 0.0140625 1422 0.225 88 0.1125 177 0.05625 355 0.028125 711 0.0140625 1422 0.225 88 0.1125 177 0.05625 355 0.028125 711 0.0140625 1422 0.225 88 0.1125 177 0.05625 355 0.028125 711 0.0140625 1422

Avg. Δy [mm]

Min Δy [mm]

4.136 2.056 1.025 0.512 0.256 8.272 4.113 2.051 1.024 0.512 12.408 6.169 3.076 1.536 0.768 16.544 8.225 4.101 2.048 1.024 20.680 10.282 5.126 2.560 1.280

3.927 1.963 0.982 0.491 0.245 7.854 3.927 1.963 0.982 0.491 11.781 5.890 2.945 1.473 0.736 15.708 7.854 3.927 1.963 0.982 19.635 9.817 4.909 2.454 1.227

Max Δy [mm] 4.430 2.220 1.111 0.556 0.278 8.860 4.439 2.222 1.112 0.556 13.289 6.659 3.333 1.667 0.834 17.719 8.878 4.444 2.223 1.112 22.149 11.098 5.555 2.779 1.390

Figure 6.23 – Image Slit Width Analysis

PDR

definition • design • development

Page 35 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.4.5. Kruse Control Kruse control is a technique by which the rotor angle can be accurately

determined without the use of an encoder at resolutions on the order of 40,000 60,000 steps per revolution. In order to accomplish this, the sense winding voltages are sampled with an A/D converter and fed back into the controller. From these values, it is possible to calculate the rotor position in real-time. The methodology is outlined below and will eventually be incorporated into this project as the controller algorithm advances.

Figure 6.24 – Two Phase Stepper Motor Schematic and Model

Vi = iR + L Vsense = L

di + Vemf dt

(6.13)

(6.14)

VA = RiA + L

diA + VemfA dt

(6.15)

VB = RiB + L

diB + VemfB dt

(6.16)

VemfA = ωR ⋅ k E ⋅ sin ( n ⋅ Θ R )

(6.17)

VemfB = ω R ⋅ k E ⋅ cos ( n ⋅ Θ R )

(6.18)

VsenseA = L

PDR

di + Vemf dt

diA + k E ⋅ ωR ⋅ sin ( n ⋅ Θ R ) dt

definition • design • development

(6.19)

Page 36 of 44

D3 Engineering VsenseB = L

Hyperspectral Camera Project (P06552)

diB + k E ⋅ ωR ⋅ cos ( n ⋅ Θ R ) dt

(6.20)

V posA = k2iA + ∫ VsenseA dt = −V pos ⋅ cos ( n ⋅ Θ R ) ∀ k2 = − k1 L

(6.21)

V posB = k2iB + ∫ VsenseB dt = V pos ⋅ sin ( n ⋅ Θ R ) ∀ k2 = −k1 L

(6.22)

V pos =

k1k E kk =− 2 E n nL

(6.23)

Figure 6.25 – Closed-loop Kruse Stepper Motor Controller

Using the same analysis performed in Figure 6.23, the relationship between the number of micro-steps per full revolution, step angle size, target size, and image slit width for the Kruse controlled system is illustrated in Figure 6.26.

PDR

definition • design • development

Page 37 of 44

D3 Engineering d [m]

y (half target width) [m]

1

0.36397023

2

0.72794047

3

1.09191070

4

1.45588094

5

1.81985117

Hyperspectral Camera Project (P06552)

Kruse Control # Slits/ μSteps / Avg. Δy Min Δy Δθ Half Full Rev. [mm] [mm] Image 40000 0.00900 2222 0.163802986 0.1570796 45000 0.00800 2500 0.145588094 0.1396263 50000 0.00720 2777 0.131065983 0.1256637 55000 0.00655 3055 0.119139193 0.1142397 60000 0.00600 3333 0.109201990 0.1047198 40000 0.00900 2222 0.327605971 0.3141593 45000 0.00800 2500 0.291176187 0.2792527 50000 0.00720 2777 0.262131966 0.2513274 55000 0.00655 3055 0.238278386 0.2284795 60000 0.00600 3333 0.218403981 0.2094395 40000 0.00900 2222 0.491408957 0.4712389 45000 0.00800 2500 0.436764281 0.4188790 50000 0.00720 2777 0.393197948 0.3769911 55000 0.00655 3055 0.357417579 0.3427192 60000 0.00600 3333 0.327605971 0.3141593 40000 0.00900 2222 0.655211943 0.6283185 45000 0.00800 2500 0.582352375 0.5585054 50000 0.00720 2777 0.524263931 0.5026548 55000 0.00655 3055 0.476556772 0.4569589 60000 0.00600 3333 0.436807962 0.4188790 40000 0.00900 2222 0.819014929 0.7853982 45000 0.00800 2500 0.727940469 0.6981317 50000 0.00720 2777 0.655329914 0.6283185 55000 0.00655 3055 0.595695964 0.5711987 60000 0.00600 3333 0.546009952 0.5235988 Figure 6.26 – Slit Width Analysis for Kruse Controlled System

Max Δy [mm] 0.1778740 0.1581152 0.1422943 0.1293622 0.1185849 0.3557479 0.3162304 0.2845886 0.2587244 0.2371698 0.5336219 0.4743456 0.4268829 0.3880866 0.3557547 0.7114959 0.6324608 0.5691772 0.5174488 0.4743396 0.8893698 0.7905761 0.7114715 0.6468111 0.5929245

6.5. Data Normalization and Material Identification The data that is captured by the imager is measured as the intensity of the pixel,

which cannot be compared directly to any reference plot of a materials spectrum. The obtained image must be converted in such a way that the resulting data set is in terms of reflectance. Several methods can be use to do such a conversion. Algorithms and methods used for this calibration and conversion will be explored during the first phases of the project development.

PDR

definition • design • development

Page 38 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.6. Mechanical System and Enclosure 6.6.1. Requirements While the some of the specifications of the project were unconstrained, the

mechanical system is to be operated in a lab environment, therefore the system should be bench top sized, but also designed to be portable if such an application is desired. Additionally, the area of hyperspectral imaging is a relatively new topic to this customer; this means maximum adjustability and functionality are of utmost importance. Due to this recent entry into the hyperspectral imaging, a tremendous amount of time is required before they can realize the cameras abilities. Lenses, controllers, motors, and the like will inevitably need to be swapped while fine tuning and hyperspectral signatures are collected. Ultimately, the camera will be assembled using parts that are precise, but unimposing to future design changes and laboratory experimentation. 6.6.2. Mounting Hardware Much of the required mounting hardware is available from the multitude

of optical equipment distributors. In some cases however, the commercially available mounting hardware isn’t adjustable enough to suit the customer’s needs. In this case, custom optical, or modified commercial equipment will be machined using the Rochester Institute of Technology’s machine shop. The camera is primarily composed of a series of lenses, a dispersing element such as a prisms or grating, a scanning mirror, and a detection device, such as an imager. 6.6.2.1. Scanning Mirror Mount The scanning mirror is a precisely controlled stepper motor with a mirror

affixed to its output shaft. To ensure that the mirror’s reflective surface is perpendicular to the ground, a mirror mount will be fabricated. The mount will allow height and angle adjustment for fine tuning. The stepper motor will be mounted to the board using an adapter plate, as the motor’s bolt pattern does not match that of the breadboard. This plate will also allow for shims should they be required.

PDR

definition • design • development

Page 39 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.6.2.2. ImSpector Lens Mounting System The ImSpector is a conveniently packaged imaging spectrograph, which is

offered in two different styles: OEM and cased. Both versions come standard with a C-mount adapter for easy interfacing to a sensor. The mount will be designed to bolt directly to the optical bench, and the ImSpector will thread into the face of the mount using the standard C-mount.

6.6.2.3. Custom Design Lens Mounts The lenses will be held using a commercially available mounting system.

Optical equipment supplier, Standa, offers a universal adjustable lens mount. This will ensure that the lenses are held perpendicular to the ground, and allows for quick lens swapping. They feature a machined flat on the bottom of the mount so they can be mounted directly to the breadboard, or to a custom mount. 6.6.2.4.Imaging Device Mount The imaging device must be held in a precise position directly in front of the

ImSpector. Finding this location will require a great deal of trial and error, so the mount must allow a great deal of adjustability. As such, the imager will be mounted on a sliding track on the ImSpector’s lens centerline. 6.6.3. System Enclosure Since the camera is going to be used in a laboratory environment, the

enclosure will simply consist of an optical breadboard and a removable aluminum top. The simple design will be functional, inexpensive, and most importantly, reliable. The optical breadboard is a perfect foundation for the camera. The flat surface and robust construction are critical for the camera’s proper operation. The evenly spaced holes will allow for an innumerable number of optical combinations. Most of the commercially available mounts are designed to be compatible with the optical breadboard, but both the mounts and the breadboard can be machined.

PDR

definition • design • development

Page 40 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

The top will be an aluminum box that will completely enclose the camera, and will be free of light leaks. It will be lightweight, and will feature a handle on top for easy removal. The top and breadboard interface will have a small layer of compressible material. This will enable the use of flexible draw-style latches. The latches are rugged, inexpensive, and forgiving if the top is not exactly aligned.

PDR

definition • design • development

Page 41 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

6.7. Preliminary Bill of Materials System/Description Imaging System Description 10 Bit 1.3 Mega Pixel CMOS Sensor DSK Camera Developer’s Kit Spectrum Digital 6416 DSK Optical System Description ImSpector Lenses Slit Prism(s) Other

Scanning Mirror Control Spectrum Digital F2812 DSP D3 Engineering Motor Control Board Mirror Motor Mechanical Components Enclosure Materials Mounting Fixtures Manufacturing Costs

Actual or Est. Cost -

Source Provided by D3 Eng. Provided by D3 Eng. Provided by D3 Eng.

Cost $2000

Source SPECIM Edmund Optics Edmund Optics Edmund Optics/Custom

Cost

Source Provided by D3 Eng. Provided by D3 Eng. Edmund Optics Provided by D3 Eng.

-

$50 Cost

Source

Figure 6.27 – Preliminary Bill of Materials

PDR

definition • design • development

Page 42 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

7. Design Feasibility and Expected Technical Issues This system is being designed to run in a laboratory environment to minimize many

of the effects that are presented when a hyperspectral camera is operated in a real-world outdoor condition. Inevitably, there are some technical issues that will be come apparent when the hardware is being evaluated and the algorithms are developed to transform the images from measured light intensity values to reflectance for identification purposes, in addition to the normalization of the measured values. Technical issues that are expected to arise are such things as separating any mixing of wavelength bands that may occur, and in the event that the camera is viewing a target at a distance in which mixing of material spectra occur, these problems will be addressed.

PDR

definition • design • development

Page 43 of 44

D3 Engineering

Hyperspectral Camera Project (P06552)

8. References [1] MicroImages Inc., Introduction to Hyperspectral Imaging. 2006 [2] "Electromagnetic Spectrum" Microsoft® Encarta® Online Encyclopedia 2005 http://encarta.msn.com © 1997-2005 Microsoft Corporation. All Rights Reserved. [3] "Spectroscopy" Microsoft® Encarta® Online Encyclopedia 2005 http://encarta.msn.com © 19972005 Microsoft Corporation. All Rights Reserved. [4] "Spectroscopy" Wikipedia, 2006 http://en.wikipedia.org. [5] ASTER Spectral Library, speclib.jpl.nasa.gov. © 1998, 1999, 2000 California Institute of Technology. [6] AVIRIS Image Cube, Jet Propulsion Laboratory, California Institute of Technology. http://aviris.jpl.nasa.gov [7] USGS Digital splib04 Spectral Library, http://speclab.cr.usgs.gov/spectral.lib04/spectrallib04.html [8] IMINT – Hyperspectral Imaging, http://www.fas.org/irp/imint/hyper.htm [9] CMOS vs. CCD and the Future of Imaging, Kodak Research and Development. http://www.kodak.com/US/en/corp/researchDevelopment/technologyFeatures/cmos.shtml. [10] Aikio, Mauri. Hyperspectral prism-grating-prism imaging spectrograph. Finland: JulkaisijaUtgivare., 2001. [11] Deardon, Steve. Spectroscopy, Astronomy, and Optics. 3 March 2004. http://astrosurf.com/dearden/ [12] Dr. Wells, Conrad. Personal interview. 11 January 2006. [13] Edmund Optics Technician. Telephone interview. 5 February 2006. [14] Hecht, Eugene and Alfred Zajac. Optics. Reading, MA: Addison-Wesley Publishing Co., 1974. [15] Hyperspectral Remote Sensing Applications, http://www.space.gc.ca/asc/eng/satellites/hyper_brochure.asp [16] Applications of Hyperspectral Imagery, http://home.student.uva.nl/derck.truijens/CPBG_files/Hyperspectral%20Applications.pdf.

PDR

definition • design • development

Page 44 of 44